Document Sample
Inaugural-Dissertation Powered By Docstoc

 zur Erlangung der Doktorwürde


      Fakultät für Physik


      Universität Bielefeld

         vorgelegt von

Diplom-Chemiker Mike Heilemann

          aus Leipzig

   Design of Single-Molecule Optical Devices: Unidirectional
             Photonic Wires and Digital Photoswitches

Gutachter:      Prof. Dr. Markus Sauer

                Priv. Doz. Dr. Andreas Hütten
Hiermit erkläre ich an Eides statt, dass ich die vorliegende Arbeit selbstständig und
ohne unerlaubte Hilfsmittel durchgeführt habe.

Bielefeld, 24. Mai 2005                                  ______________________
                                                                Mike Heilemann

Molecular photonics is a new emerging field of research around the premise that it is
possible to develop optical devices using single molecules as building blocks.
Currently used waveguides, applied for example in telecommunication, rely on the
classical physics of bulk materials: Maxwell’s equations allow propagating modes in
the far field, and the wavelength of light imposes a fundamental lower limit on device
size. However, nature has evolved several examples of photonic nanostructures to
guide light over much smaller length scales for “light harvesting” in plants and
photosynthetic bacteria. This fundamentally quantum mechanical solution is most
often based on near-field dipole-dipole interactions, i.e. fluorescence resonance
energy transfer (FRET). As a consequence, light-harvesting complexes, one of
nature’s supreme examples of nanoscale engineering, have inspired researchers to
engineer molecular optical devices, such as molecular photoswitches or molecular
photonic wires. A molecular photonic wire is distinguished from a molecular
electronic wire by supporting excited-state energy transfer rather than electron- (or
hole-) transfer processes and could find application in, for example, optical
computing as short-range interconnects in dense optical circuits. The excited state
resulting from light absorption by one chromophore migrates among an array of
chromophores, ultimately reaching a fluorescence dye to output an optical signal. In
1994, Lindsey and co-workers realized the first molecular photonic wire based on
conjugated porphyrin arrays. However, strong coupling in porphyrin arrays has the
disadvantage of forming so-called energy sinks due to different local interactions of
the chromophores. As a further prerequisite, molecular photonic wires have to
operate at the single molecule level and because such complex molecular systems
are expected to exhibit a high degree of heterogeneity they have to be characterised
at an individual basis as well.

In this work an alternative access to molecular photonic wires was elaborated. This
approach was based on (i) the use of conventional, single molecule compatible
chromophores, (ii) an energy cascade as the driving force for the excited-state
energy to ensure unidirectionality, and (iii) an arrangement of chromophores such
that strong electronic interactions promoting fluorescence quenching are prevented.


In the course of this work, the following challenges had to be met: (i) selection of
suitable chromophores, (ii) development of a chemistry to obtain a regular
arrangement of chromophores which allows for efficient energy transfer but prevents
alterations of photophysical properties resulting in quenching, (iii) development of a
single molecule set-up which allows investigation of the performance of individual
photonic wires, (iv) elaboration of optimal conditions for the investigation and
functioning of individual photonic wires and (v) development of a molecular optical
switching unit.

The main requirement the chromophores have to fulfil is their compatibility with single
molecule spectroscopy. This affects in particular their photostability, their
photophysics and possible ways to improve their performance by optimising
conditions. Therefore, more than 20 chromophores of different classes (i.e.
rhodamines, oxazines and carbocyanines) were investigated under various
conditions. Strikingly, chromophores belonging to the same class showed
comparable behaviour with respect to their reactions to different conditions, such as
oxygen concentration or reducing and oxidizing agents. For example, oxazine
derivatives showed longest survival times and minimal photophysics such as blinking
under ambient conditions in buffer. Rhodamines’ photostability, on the other hand
could be increased by more than tenfold upon addition of reducing agents such as β-
mercaptoethanol (MEA), while strong blinking due to long triplet states or charge-
separated states was observed upon oxygen removal. Carbocyanines required both,
removal of oxygen to increase photostability plus addition of MEA, which here acts as
triplet quencher and reduces blinking. This comparative study showed that a photonic
wire should be constructed from only one class of chromophores as no conditions
could be found satisfying the requirement of all classes. As rhodamines offer the
broadest spectral range of chromophores, they were most frequently used in the
photonic wires developed.

To achieve a very regular arrangement of chromophores, DNA was used as rigid
scaffold. The well-developed labelling and post-labelling strategies of DNA were
exploited to introduce a variety of different chromophores in a modular conception. It
was shown that best results could be obtained when dye-labelled oligonucleotides
were hybridised against a long DNA-strand already carrying a primary donor
chromophore and a biotin for specific immobilization. The distance between
subsequent chromophores was adjusted to 3.4 nm, i.e. 10 bases, which ensured

efficient FRET and prevented direct orbital interaction. Photonic wires were
synthesised carrying up to 5 chromophores and covering a spectral range from 488
nm to 750 nm. In ensemble experiments the maximum overall transfer efficiency was
determined to be 21%. However, as indicated by steady state and time-resolved
measurements, a broad heterogeneity within the samples was suspected. To
disentangle the complexity of the photophysics of so-built photonic wires, a novel
fluorescence microscope, single molecule sensitive on four spectrally separated
detectors, was developed. The confocal set-up was operated with multiple laser
excitation wavelengths and offered the possibility for time-resolved fluorescence
experiments, two-colour applications, and polarization-modulated excitation.

For the first time, a quadruple jump of energy transfer along a single photonic wire
containing five chromophores and adsorbed on a glass surface was demonstrated
with an overall transfer efficiency of ~90%. Confirmation that the energy is transferred
stepwise comes from prolonged excitation of single molecules, which results in
sequential photobleaching and a shift in the emission from the red back towards the
blue. Furthermore, collective transitions of whole photonic wire molecules into
nonfluorescent dark states were observed. It was demonstrated that fluorescence
spectra from a large number of single photonic wire molecules resembled the
ensemble spectrum of the sample.

To increase the homogeneity of photonic wires and better control the photostability,
the molecules were anchored on a protein surface by biotin/streptavidin binding in an
aqueous environment. Best stability and long observation times of single constructs
were attained by using four rhodamines. Here, energy transfer efficiencies of up to
~90% were observed. Photonic wires with five fluorophores in aqueous solution used
carbocyanine and carbopyronine derivatives as the final emitting unit, since no long-
wavelength absorbing and emitting rhodamines are available for conjugation
chemistry. Fluorescence lifetime information revealed further aspects of energy
transfer, and complemented spectral data in order to identify fluorophores involved in
particular energy transfer steps. Leakages in energy transfer, created by
photodestruction of a fluorophore inside the chain, were revealed. Polarization
modulation of the excitation light in combination with fluorescence lifetime gave
insight into the rotational mobility of the fluorophore serving as input unit, i.e.
Rhodamine Green. Three subpopulations, differing in quantum yield, fluorescence
lifetime, and degree of rotational freedom, were found.

To further improve the performance of DNA-based photonic wires, a method of
subsequent hybridisation of oligonucleotides to an immobilized single-stranded DNA
was developed. This highly-efficient process was traced at the single-molecule level
and yielded up to 90% of a desired target DNA within minutes. By this technique, (i)
sample heterogeneities from ensemble hybridisation were reduced to a minimum and
(ii) hindered hybridisation was observed for one hybridisation step, requiring longer
incubation times. This observation can be attributed to a less favourable
conformation or secondary structures of oligonucleotides, and explains relatively low
ensemble energy transfer efficiencies measured in photonic wires with five
fluorophores. A fraction of ~30% of all molecules showed energy transfer efficiencies
with ~70%, in agreement with experiments carried out on dry glass substrates.
Furthermore, single-molecule hybridisation represents a striking tool for a stepwise
construction of complex geometrical arrangements to overcome kinetic hindrances.

After the accomplishment of the photonic wire, a further goal was the development of
a molecular photoswitch. Hitherto, only one demonstration of chemically synthesized
photoswitching of single molecules at room temperature had been reported. In the
context of this work, it was shown that commercially available unmodified
carbocyanine dyes such as Cy5 and Alexa647 could be used as efficient reversible
single-molecule   optical   switch,   whose     fluorescent   state   after   apparent
photobleaching can be restored at room temperature upon irradiation in the range of
488 – 532 nm. In oxygen-free environment and in the presence of 100 mM β-
mercaptoethanol (MEA) as triplet quencher, more than 20 switching cycles could be
achieved for single Cy5 molecules with a reliability of >90%. To further characterize
the photophysical properties of the reversible switchable state, an energy transfer
donor, TMR, in proximity of Cy5 was used to report on the “off” states of the acceptor
Cy5. Examination of the single pair FRET (sp-FRET) with high time resolution
revealed the existence of three intermediates prior to fluorescence restoration. In
addition to the importance of such single-molecule photoswitches e.g. for optical data
storage, the results presented in this work imply limitations for the use of
carbocyanine dyes in sp-FRET experiments.

Table of Contents

SUMMARY .................................................................................................................. I

TABLE OF CONTENTS ............................................................................................. V

1. INTRODUCTION .................................................................................................... 1

2. THEORY............................................................................................................... 11

2.1. Fluorescence Microscopy............................................................................... 11
   2.1.1. Basic Principles of Fluorescence ................................................................ 12
   2.1.2. Molecular Interactions Influencing Fluorescence ........................................ 25
   2.1.3. The Role of Oxygen .................................................................................... 33
   2.1.4. Fluorescent Probes ..................................................................................... 36

2.2. Single-Molecule Spectroscopy....................................................................... 44
   2.2.1. Confocal Microscopy................................................................................... 45
   2.2.2. Confocal Microscopy at the Single-Molecule Level ..................................... 50
   2.2.3. Single-Molecule Intensity Fluctuations ........................................................ 51
   2.2.4. Single-Molecule FRET (smFRET)............................................................... 57

3. MATERIALS AND METHODS ............................................................................. 61

3.1. Spectrally-Resolved Fluorescence Lifetime Microscopy............................. 61
   3.1.1. Time-Resolved Fluorescence Microscopy .................................................. 71
   3.1.2. Time-Correlated Measurements at the MicroTime 100 Set-up ................... 78
   3.1.3. Two-Colour Excitation ................................................................................. 80

3.2. Ensemble Spectroscopic Instrumentation .................................................... 81
   3.2.1. Steady-State Measurements....................................................................... 81
   3.2.2. Time-Resolved Spectroscopy ..................................................................... 82

3.3. Biological and Chemical Methods ................................................................. 84
   3.3.1. Conjugation Chemistry................................................................................ 84
   3.3.2. Surface Preparation Techniques................................................................. 87

                                            Table of Contents

  3.3.3. Influencing the Chemical Environment of Fluorophores.............................. 89

4. RESULTS AND DISCUSSION............................................................................. 93

4.1. Photophysical Properties of Fluorescent Dyes ............................................ 93
  4.1.1. Three Classes of Dyes: Carbocyanines, Rhodamines and Oxazines ......... 94
  4.1.2. Cy5 at the Single-Molecule Level.............................................................. 100
  4.1.3. MR121 at the Single-Molecule Level......................................................... 104
  4.1.4. ATTO647 at the Single-Molecule Level..................................................... 108
  4.1.5. Rhodamine Green at the Single-Molecule Level....................................... 112
  4.1.6. Rhodamines in smFRET-Pairs at the Single-Molecule Level.................... 119

4.2. Design of a Unidirectional Photonic Wire ................................................... 124
  4.2.1. Configuration of Photonic Wires Based on 60bp DNA .............................. 126
  4.2.2. Estimating Energy Transfer Efficiencies in Photonic Wires from Ensemble
  Steady-State Measurements............................................................................... 129
  4.2.3. Time-Resolved Ensemble Spectroscopy of Photonic Wires ..................... 138

4.3. Studying Photonic Wires with Single-Molecule Spectroscopy ................. 145
  4.3.1. Spectrally Resolving Single Photons on Four Detector Channels............. 146
  4.3.2. Photonic Wires Adsorbed on Dry Glass Substrate.................................... 148
  4.3.3. Photonic Wires Immobilized in Aqueous Solution ..................................... 154
  4.3.4. Building-Up Single Photonic Wire Molecules on a Surface ....................... 159
  4.3.5. Time-Resolved Single-Molecule Spectroscopy of Photonic Wires ............ 162
  4.3.6. Polarized Excitation of the Input Unit Rhodamine Green .......................... 169

4.4. Carbocyanine Dyes as Optical Single-Molecule Switch............................. 172
  4.4.1. Switching of Single Cy5 Molecules Immobilized in Solution...................... 173
  4.4.2. Mechanistic Studies on Cy5-Photoswitch ................................................. 175
  4.4.3. Optical Switching of Cyanine Dyes in Ensemble Experiments.................. 181

5. CONCLUSION AND OUTLOOK........................................................................ 183

5.1. DNA Based Photonic Wires .......................................................................... 183

5.2. Single-Molecule Photoswitch: A Mechanistic View.................................... 189

                                                Table of Contents

6. REFERENCES ................................................................................................... 195

7. PUBLICATION LIST .......................................................................................... 211

7.1. Publications in Scientific Journals .............................................................. 211

7.2. Conference Presentations ............................................................................ 212

8. ABBREVIATIONS .............................................................................................. 213

9. ACKNOWLEDGEMENTS .................................................................................. 215

Table of Contents

1. Introduction

Nanometer scale optical architectures are of great interest as photonic and electronic
devices with potential applications in dense optical circuits, optical data storage and
materials chemistry [Yablonovitch, 2001; Vukusic and Sambles, 2003; Hu and
Schulten, 1997]. Two prominent examples of optically addressable nanostructures
which were the subject of this work represent molecular photonic wires and optical

While classical optical waveguides rely on propagating modes in the far field,
nanometer-sized molecular photonic devices guide light via near-field interactions of
molecules in close proximity. In other words, molecular photonic wires transfer light
via electronic excitation transfer (EET). On the level of nanometer-sized molecular
devices, the transport of excitation energy is advantageous because it circumvents
the connection problem present in electric wires, i.e. the bottleneck that occurs when
trying to connect molecular devices with macroscopic ones. In the case of a
molecular photonic wire, excited state energy is induced into an input unit by means
of light, transported through transmission elements and finally emitted at another
wavelength and location by an output unit (see figure 1-1). Otherwise, the energy can
be used for an electron-transfer reaction, i.e. the conversion of excited-state energy
into an electric charge with the possibility for subsequent chemical reactions.

Long before the design and construction of efficient photonic wires with molecular

      Figure 1-1: Working principle of an electronic wire compared to a photonic wire.


dimensions were subject in research, nature had already evolved several examples
of photonic nanostructures to guide light from the light-harvesting complexes to the
reaction centre as the initial steps of photosynthesis [Deisenhofer et al., 1984;
Glazer, 1989; Hu and Schulten, 1997; Glazer, 1989] (illustrated in figure 1-2). After
absorption of a photon by a pigment-protein complex has occurred, the excitation
energy is conveyed from one light-harvesting chromophore to another in a series of
radiationless transfers, which end at a special pair of chlorophyll molecules within the
transmembrane reaction centre complex. Often, hundred to thousands of pigment
molecules are associated with one reaction centre, and the energy transfer from the
absorbing pigment to the reaction centre can comprise up to hundreds of energy-
transfer steps.

Intensive experimental and theoretical efforts have been ventured to understand the
energy-transport mechanisms in such natural light-harvesting complexes. In
particular, the large energy-transfer efficiency achieved by these antenna complexes
has stimulated the synthesis of various artificial multichromophoric systems to mimic
natural photosynthetic light harvesting systems. Most approaches to artificial systems
synthesised arrays of covalently linked chromophores with a specific design to
ensure large collection efficiencies and fast and efficient energy migration. Like the
natural antenna, most of these systems are based on porphyrin pigments [Wagner

   Figure 1-2: Arrangement of a light harvesting complex in the photosynthetic reaction
   centre of bacteria (left; subunits LH-I and LH-II, reaction centre RC). LH-II complexes
   (right side) are responsible for energy transfer to LH-I and serve as antenna complexes.
   (images are courtesy of the Theoretical Biophysics Group, University of Illinois, Urbana-


and Lindsey, 1994; Seth et al., 1996]. Lindsey and co-workers first demonstrated in
1994 that a wire composed of a boron-dipyrrin (BDPY) input unit, three zinc
porphyrins and a free-base porphyrin (Fb) joined via diarylethyne linkers shows
predominately emission from Fb after excitation of BDPY at 488 nm. The end-to-end
energy-transfer efficiency in the wire was estimated to be 76%.

However, a conceptual difference between light-harvesting complexes and photonic
wires has to be pointed out: the former usually exhibit dendritelike structures with the
goal being to rapidly and efficiently transfer energy to a reaction centre, and to
operate the reaction centre at full capacity. Light-harvesting antennae are exemplary
of efficient energy transfer. On the other hand, they are not optimised to transport
excited-state energy unidirectionally over long distances.

The key parameters for the design of an artificial EET system are the selection of
suitable fluorophores and the precise control of their interactions. Nature uses both
parameters to optimise the absorption efficiency and the energy flow from the
harvesting complexes to the reaction centre. The pigments used exhibit an energy
cascade, which directs the energy towards the lower energy sites and thereby
towards the reaction centre. The arrangements of pigments is such that the energy is
quickly and efficiently transferred between chromophores without loosing energy by
alternative pathways, such as internal conversion or energy transfer processes. It is
interesting to note that, in order to fulfil this task, not a single energy transfer
mechanism is employed, but depending on the species and location in the energy
transfer cascade, different mechanisms are realized.

Firstly, chromophore interactions have to be divided into two regimes depending on
the extent of their electronic interaction: i) In some cases, the chromophores are
arranged so closely that their wave functions mix strongly to produce new,
delocalised states (exciton states) as, for example, in the B850 ring of LH2 or in
chlorosomes [van Oijen et al., 1999; Psencik et al., 2003]. The presence of excitonic
states with different energies leads to substantial changes in the absorption spectrum
of the “supermolecule” compared to that of the composite of the individual
chromophores (figure 1-3 b). In this strong coupling regime, the interaction energy is
much larger than the vibrational energy, so that numerous energy transfers can take
place during a single vibration. Under these conditions, electronic excitation becomes
a “communal” phenomenon, and intramolecular vibrations are uncoupled from


 Figure 1-3: (a) Schematic working principle of a photonic wire: light is collected by a
 funnel-like input unit and transferred via EET to an output unit. Interactions of fluorophores
 can be divided into: (b) In the strong coupling limit (coherent EET), donor and acceptor
 electronic states mix strongly, which results in delocalisation of the excitation energy
 ideally over the whole molecule. In the weak coupling limit, fast nuclear relaxation localises
 the initial excitation on the chromophores prior to stepwise EET via a cascade of
 chromophores (c) or via energy hopping (d).

electronic excitation, with far-reaching changes in the shape of the absorption band.
ii) In most cases, however, the chromophores are spaced further apart and show
weak electronic interaction. Here, energy transfer is governed by Coulombic
interactions, and the absorption spectrum constitutes the sum of the individual
components. In this weak coupling regime, fast nuclear relaxation localises the initial
excitation prior to EET. Accordingly, EET can be well-described within the framework
of the FÖRSTER theory [Förster, 1948]. Coulombic interactions can occur either
between a donor-acceptor pair with distinct absorption and fluorescence spectra
(figure 1-3 c), or between identical chromophores, if they exhibit sufficient energetic
overlap for this process (figure 1-3 d). This so-called homotransfer or energy hopping
represents a key mechanism for energy transport in some light-harvesting
complexes. The challenge in constructing a photonic wire that is optimised for long-
range energy transfer is the compromise between providing directionality by
introducing an energy cascade and minimising energy loss by exploiting energy
hopping as energy-transfer mechanism.

As an extension to mere energy transfer over longer distances in artificial molecular
photonic wires, it is of great interest to control the energy flow, i.e. by applying a
switching unit. Over the past few years, molecular switches of different types have


been intensively researched, in the quest for molecular electronic devices. Some
switching systems operate by a conformational change in the molecule, induced by
either an electric field [Donhauser et al., 2001], a STM (scanning tunneling
microscope) tip [Moresco et al., 2001], an electrochemical reaction [Bissell et al.,
1994] or light [Giordano et al., 2002]. Alternatively, molecular switches can be
operated nonconformationally by redox reactions [Gittins et al., 2000] or a chemical
binding event [Kasibhatla et al., 2003]. Very recently, it was demonstrated that the
fluorescence of individual dye-labelled DNA molecules can be reversibly switched
from green to red and vice versa upon application of an electric field [White et al.,
2004]. In the present work, the focus was set onto a combined chemical and light-
induced manipulation of intermediate states of single fluorescent molecules.

It is the complexity of the described systems - independent of the energy-transfer
mechanism employed - which determines the need for new analytical techniques for
the characterisation of bottom-up nanotechnological devices, such as photonic wires
and photoswitches. Single-molecule fluorescence spectroscopy (SMFS) is a
technique that provides detailed information required for the analysis of static
heterogeneity [Kapanidis et al., 2005]. In addition, SMFS also enables to probe the
quality of the device. Molecular photonic wires and photoswitches have to operate at
the single-molecule level and, hence, they have to be characterised at this individual
level as well.

Optical single-molecule detection was first realized in the year 1976 with the
detection of single antibodies labelled with 80 to 100 fluorophores [Hirschfeld, 1976].
Thirteen years later, two groups independently demonstrated the detection of a
single fluorescent molecule at cryogenic temperature in a solid host matrix [Moerner
and Kador, 1989; Orrit and Bernard, 1990]. A short time later fluorescence emission
from single molecules could be detected at room-temperature [Shera et al., 1990]. An
important step towards better sensitivity in SMFS was made in 1992 by combining
this technique with the principle of confocal microscopy [Rigler et al., 1992]. In the
following years, a dramatic increase of research efforts towards new techniques
based around SMFS were observed.

Nowadays, confocal SMFS is a well-established technique, which is elaborated in a
number of reviews [Nie and Zare, 1997; Moerner and Orrit, 1999; Ambrose et al.,
1999; Boehmer and Enderlein, 2003; Moerner and Fromm, 2003; Tinnefeld and


Sauer, 2005; Neuweiler and Sauer, 2005]. Applications of this technique touch
various fields of research, comprising physical, chemical and biological sciences
[Weiss, 1999; Bustamante, 2000; Weiss, 2000]. The strength of this technique lies in
the ability to detect single fluorophores both in solution and adsorbed onto a surface.
Averaging    of   molecular    properties   observed     in   ensemble     experiments   is
circumvented, and subpopulations or molecular processes otherwise not accessible
can be investigated using SMFS.

As an important rule underlying the observation of single molecules, the ergodic
principle concludes that the sum of many single-molecule events, integrated over a
long period of time, should reflect the results from an ensemble measurement. In
other words, the result of an average measurement is also obtained from a single
molecule, observed for a long time. The principle is portrayed in figure 1-4: if a single
fluorophore is observed, a distinct wavelength of each emitted photon can be
recorded. All photons emitted by this molecule can be summed up and represent the
emission spectrum of an ensemble of fluorophores.

 Figure 1-4: Representation of the ergodic printiple: the emission wavelength of
 fluorescence photons from one single fluorophore observed for a long time (left side)
 reflects the averaged emission spectrum of an ensemble of fluorophores (right side).

Due to a very small excitation volume in confocal microscopy, SMFS in solution can
be used for very sensitive detection of fluorescent probes down to concentrations of
10-12 M, which also opens applications in diagnostic research [Neuweiler et al., 2002].
Adsorbed onto a surface, the fluorescence response of a single fluorophore and its


                                                    Figure 1-5: Confocal fluorescence image of
                                                    fluorophores anchored on a protein surface
                                                    (10 µm x 10 µm). Fluorescence signatures of
                                                    single molecules exhibit “blinking” and

interactions with the nearest environment can serve as a probe for the nanoscopic
environment [Macklin et al., 1996].

The main technique for single-molecule detection applied in this work is spectrally-
resolved fluorescence lifetime imaging microscopy (SFLIM) [Tinnefeld et al., 2001]. A
confocal line scanning microscope was equipped with four spectrally separated
detectors and various excitation sources. The method goes beyond the mere
detection of fluorescence photons and provides additional information, such as
fluorescence lifetime and emission wavelength. Together with the modulation of the
polarization of an excitation light source, the method provides a powerful tool for the
investigation of multichromophoric compounds. Fluorescence signatures derived
from single fluorophores anchored on a protein surface are portrayed in figure 1-5.
The scan image shows two typically observed properties of single fluorophores,
which are intermittencies in fluorescence emission (often termed “blinking”) due to
reversible   transitions   into   dark-states       of   different   nature   and   irreversible
photodestruction (“photobleaching”).

To be able to investigate the performance of complex nanomolecular constructs at
the single-molecule level, the device has to fulfil certain conditions in terms of
photostability and fluorescence quantum yield. Fluorophores must be selected
carefully, and conditions for high photostability have to be found for a set of
fluorophores to realize multichromophoric compounds. Combined with a modular
approach, the construction of complex devices is simplified. This underscores the
strength of the DNA-based conception, which enables the free choice of fluorophores
and manifold arrangements and combinations of fluorophores.


Recent developments in nanotechnology and optoelectronics have focused research
attention onto the possibility to use single fluorescent molecules as molecular
photonic switches and optical data storage elements [Gittins et al., 2000; Dickson et
al., 1997; Irie et al., 2002]. To store one bit per molecule by its fluorescence intensity
in a reversible fashion, single molecules have to be switched digitally in a controlled
manner by external stimuli. Usually, changes in fluorescence intensity from single
molecules are attributed to quenching, stochastic intersystem crossing events to
triplet states, or spectral diffusion due to fluctuations in the local environment of the
chromophores. In cases where such fluctuations can be controlled, highly
reproducible switching can be achieved, as was shown for light-induced frequency
jumps in liquid helium temperature experiments [Basché and Moerner, 1992]. The
green fluorescent protein (GFP) and some derivatives constitute the first room
temperature all-optical examples of chromophores that can be reversibly switched
between different nonfluorescent and fluorescent states at the single-molecule level
[Dickson et al. 1997; Peterman et al., 1999; Jung et al., 2005, Chirico et al., 2004].
These natural photoactivatable chromophores are particularly interesting for precise
photolabelling and tracking of proteins in living cells [Chudakov et al., 2004].

More recently, the first room temperature single-molecule photoswitch based on
optical switching of the transfer efficiency in a fluorescence resonance energy
transfer (FRET) pair was demonstrated [Irie et al., 2002; Fukaminato et al., 2004]. In
a two-color experiment, a donor chromophore (bis(phenylethynyl)anthracene)
connected to a switchable quenching unit (a diarylethene derivative) could be
switched on and off by 488- and 325-nm light, respectively. UV light was used to
activate the quencher (energy transfer acceptor), while 488-nm light was used for
deactivation of the quenching unit and probing of the fluorescence of the donor
chromophore. The use of identical wavelengths (488 nm) for probing and switching
was possible because the deactivation (isomerisation) is about 1000 times less
efficient than the activation of the quenching unit. Thus, probing and isomerisation
can be controlled by changing the excitation light intensity.

On the other hand, single-molecule fluorescence experiments have revealed several
expected and unexpected photophysical phenomena of the carbocyanine dye Cy5
such as cis-trans isomerisation, off states additional to triplet formation, and complex
photobleaching pathways including nonfluorescent intermediates that still absorb light
in the visible range [Ha 1999; Tinnefeld 2001; Tinnefeld 2003; Ha 2003; Widengren

and Schwille 2000]. These facts raise hope that a controlled microenvironment might
stabilize intermediate states and open the way for reversible transitions, recovering
the fluorescent state of the fluorophore.

This thesis presents the design and spectroscopic investigation of photonic
nanostructures such as unidirectional molecular photonic wires and optical switches.
Methods involved to scrutinise the working principle of these optical elements include
ensemble spectroscopic and single-molecule fluorescence techniques. In the frame
of this work, a confocal line scanning microscope with four detector channels and
various excitation sources for laser-induced fluorescence was built. The set-up
allowed time- and spectrally-resolved detection of single fluorescence photons and
modulation of the polarization of the excitation light. The sum of information obtained
from each detected photon qualified the single-molecule set-up for spectroscopic
studies on complex multichromophoric systems.

The main focus was on photophysical studies of unidirectional molecular photonic
wires and digital photoswitches. The synthetic strategy to design photonic wire
structures involved DNA as a rigid scaffold and molecular building block system.
Light energy was injected into an input unit, transported along a multichromophoric
arrangement of fluorophores via dipole-dipole energy transfer, and emitted by a final
unit. Suitable fluorophores were attached chemically to single oligonucleotides, and
hybridisation allowed the construction of different arrangements of fluorophores.
Strategies towards reduced heterogeneity, improved photostability and controlled
photophysics were elaborated. Furthermore, stabilizing conditions for the chemical
class of rhodamine derivatives were investigated and optimised.

A second focus was set onto the development of another photonic device, i.e. an
optical switching unit constituted of a fluorophore. An optical photoswitch was
realized on the basis of commercially available carbocyanine dyes, e.g. Cy5 (Roche,
USA) and Alexa647 (Molecular Probes, USA). An understanding of the complex
pathway of photobleaching, exhibiting many intermediate states which can selectively
be manipulated, allowed the microenvironment of individual carbocyanines to be
arranged in a way that enabled the optical generation of reversible transitions into a
dark state. By optical manipulation with two different excitation wavelengths, highly
reversible “switching” of fluorescence could be demonstrated.


2. Theory

2.1. Fluorescence Microscopy

In the last decades, a trend towards working interdisciplinary between the classical
fields of science has led to the emergence of new principal techniques. One widely
used technique is fluorescence microscopy, which has developed into a method
widely used in many different fields of research.

Fluorescence was first observed by SIR GEORGE G. STOKES in the middle of the
nineteenth century. He made the observation that the mineral fluorspar showed blue
light emission when illuminated with ultraviolet light, and coined the word
"fluorescence". Stokes observed that fluorescence exhibits longer wavelengths than
the excitation light, a phenomenon that has become known as the STOKES-shift.

Nowadays, fluorescence microscopy has developed into a widely applied tool in
many fields of research ranging from molecular biology and biochemistry to
chemistry and physics [Lakowicz, 1999]. Any system that is fluorescent or can be
modified in a way that it becomes fluorescent is suitable for fluorescence microscopy.
Therefore, the technique profits enormously from a large number of fluorescent
probes, e.g. organic chromophores, nanocrystals or quantum dots, metallic clusters
and fusion proteins. It allows the investigation of processes on a large both temporal
and spatial scale, ranging from nanoseconds to seconds, and nanometers to
micrometers. The spectrum of uses includes colocalization of biological substrates in
cell compartments using widefield mercury lamp excitation, to single-molecule
sensitive high-precision colocalization with few nanometer accuracy by spectrally
resolved fluorescence lifetime microscopy (SFLIM) [Heilemann et al., 2002]. Beyond
basic fluorescence information, a large number of additional parameters, e.g.
spectral characteristics, polarization or coincidence, allow a high level of description
in an observed system [Heinlein et al., 2005].

Fluorescence microscopy is the main technique used in this work. The basic principle
of fluorescence, the spontaneous emission of a photon upon electronic excitation of a
molecule via light absorption, shall therefore be described in detail in the following



The interaction between an electromagnetic wave with matter, i.e. atoms or
molecules, is based on resonant coupling of an incoming light wave inducing
oscillations in a second system [Atkins, 1995]. Depending on the energy of the
electromagnetic wave and the nature of the coupled system, the excitation of
rotational, vibrational or electronic states may be induced. For any process of
interaction between light and matter, the frequency condition of NILS BOHR represents
the link between light frequency υ and transition energy ∆E ,

                               hυ = ∆E = E0 − E1                               ( 2-1 )

where h is PLANCK’s constant. If the resonance condition is satisfied and the energy
of the incoming light is suitable to excite an atom or molecule from a lower energy
level E0 to a higher energy level E1, the strength of an interaction between an
electron and the electric field E is related directly to the ability of the electron to
“follow” the light wave and to the magnitude of the maximal charge separation
effected by this interaction. The magnitude of development of charge separation as
one proceeds from a ground state, represented by the wave function Ψ0, to an
excited state, represented by the wave function Ψ1, is related to the transition dipole
moment µ 0→1 ,

                                µ0→1 = ∫ Ψ1*µΨ0
                                            ˆ                                  ( 2-2 )

where µ is the operator of the electric dipole moment [Atkins, 1995]. As a
fundamental requirement for absorption and emission, the value of µ 0→1 must be

finite. Described with other words, a transition can interact much better with the
electric field if the charge rearrangement has an explicit dipole character.

The transition of a molecule from a lower energy state to a higher energy state is
explained by oscillations between the electromagnetic field and the frequency of a
transition. This process is described as induced absorption and depends on the


energy density of the electromagnetic field, ρ . The transition probability w, which
describes the change of the probability to find a molecule in an excited state,          ,
has been described by EINSTEIN to be

                                  w=      = Bρ                                    ( 2-3 )

In this equation, B is the EINSTEIN coefficient for induced absorption,
                                        r     2
                                      µ 0→1
                                   B=                                             ( 2-4 )
                                      6ε 0 h 2
showing a square dependency on the value of µ 0→1 ( ε 0 is the dielectric constant in

the vacuum). EINSTEIN could show that the coefficient for induced emission, a
process important for generation of laser light, equals the coefficient for induced

For any further description, we must now differentiate between an isolated atom, or a
molecule consisting of several atoms. In the first case, only electronic transitions are
possible, and the degrees of freedom for motion are limited to translation of the atom.
As a result, electronic transitions in atoms are discrete, or, in other words, atomic
spectra are line spectra. Depending on the type of atom and the shell the excited
electron originated from, excitation energies lie in a range of less than 1 eV up to
hundreds of eV. Transitions in the visible light region are typically in the range of 1 to
4 eV and are usually attributed to valence band electrons.

In the case of a molecule, e.g. an aromatic compound, a certain number of degrees
of freedom for both vibrational and rotational transitions exist, additional to electronic
transitions. At a first point, this means that an excitation will not occur purely of
electronic, vibrational or rotational nature only, but as a mixture out of all. Reflecting
the energy gap between the states themselves, they are energetically separated by a
factor of 100. If using wavenumbers to express the energetic difference, a common
habitude in spectroscopy, characteristic values of around 10 cm-1 for rotational,
around 1000 cm-1 for vibrational and around 100 000 cm-1 (which equals a few eV)
for electronic transitions can be given. The important point here is that any light
induced transition of a molecule, following the absorption of a photon with
appropriate energy, leads to a mixed excitation of different nature. Different to atoms,


these transitions are broader due to the mixing of different states, leading to broader
band spectra for molecules in general.

Furthermore, we have to distinguish between the observation of absorption and
emission of molecules. A chromophore is defined with respect to a molecule’s
absorption properties, whereas a lumophore describes a molecule exhibiting light
emission. If emission is caused by fluorescence, the term fluorophore is used.

In the following general considerations for molecules, rotational levels are excluded
for clarity, and only electronic and vibrational transitions are taken into account. In
this context, a simultaneous transition of a mixed nature of both vibrational and
electronic kind is often termed vibronic.

Light absorption is a process which
can be regarded as instantaneous,
and transitions occur at a time scale of
10-15 s. Since the mass of an electron
is at least three orders of magnitude
lower than the mass of the nuclei, the
transition time is too short for any
significant displacement of nuclei. As a
consequence, all electronic transitions
in the energy-distance plot are vertical,
which is described by the FRANCK-
CONDON     principle   [Condon     1928]
                                              Figure 2-1: FRANCK-CONDON-principle: Excitation
(figure 2-1). In more detail, this means
                                              of electronic transitions is fast with respect to
that an electronic transition out of the      nuclei movement, which leads to horizontal
lowest vibrational state of the ground        transitions into excited vibrational states.
state S0 of the molecule, which is
mainly populated at room temperature conditions, will take place into a higher
vibrational level of the first electronic excited state. Upon this excitation, the
molecules rapidly relax to the lowest vibrational energy of the first excited electronic
state, S1, a process referred to as vibrational relaxation. Spontaneous emission of a
fluorescence photon with a similar time scale as light absorption follows the FRANCK-
CONDON principle, which goes hand in hand with a vertical transition from the lowest
vibrational level of S1 to a higher vibrational energy in S0. As a result, if a particular


transition probability, also known as FRANCK-CONDON factor, between the first and
second vibrational level is largest in absorption, the reciprocal transition is also most
probable in emission. This fact explains the mirror image rule of fluorescence, which
shall be discussed later.

A more detailed view of all possible processes in an energetic scheme of a molecule
can be given using the JABLONSKI diagram, and a typical example is shown in figure
2-2. Here, the singlet ground, first and second electronic states are depicted by S0,
S1 and S2 and so on. Triplet states of the molecule are depicted T1, T2 and Tn. At
each electronic level, numerous vibrational and rotational levels exist, where a
molecule can exist for a certain time. To simplify the scheme, only vibrational levels
are included.

    Figure 2-2: JABLONSKI-diagram, showing a simplified scheme of energetic levels
    for organic chromophores (rotational states are neglected). ISC: intersystem
    crossing, VR: vibrational relaxation, IC: internal conversion.

Depending on the transition probabilities of a given chromophore, the interaction of
an electron with the electric field may excite a molecule out of the ground state S0 to
any vibrational excited state of a higher electronic singlet state Sn. In a next step,
molecules usually relax to the higher vibrational state of S1 by internal conversion on


a sub-picosecond timescale, followed by vibrational relaxation, occurring in ~10-12 s.
The energy is hereby dissipated as heat via collisions with neighbouring molecules.
Compared to the natural lifetime of the first excited state S1 being around 10-8-10-9 s,
all internal conversion processes are usually complete before emission of a
fluorescence photon.

From the first excited state S1, several radiative and non-radiative pathways of
depopulation are now possible for a molecule to return to the ground state S0, which
are either directly or indirectly. The one leading to fluorescence is the radiative
depopulation of S1 by spontaneous emission of a photon. According to the FRANCK-
CONDON principle, this process is described by a vertical transition to a higher excited
vibrational level of the ground state S0, followed by vibrational relaxation again, and
hereby reaching thermal equilibrium.

The non-radiative depopulation process of S1 is coined internal conversion (IC),
which can be described as a close approach between the energy landscapes of both
S1 and S0, allowing electrons to “tunnel” between them. As a result, the molecule will
be found in a highly excited vibrational level of the ground state, which is deactivated
by vibrational relaxation.

A molecule in the S1 state can also undergo a spin conversion to the first triplet state,
T1. Since this transition is not spin allowed, these events are rare, and kinetic rates
strongly depend on the nature of a chromophore and the transition probability. From
this triplet state T1, similar processes as already discussed for the first excited singlet
state S1 are now possible, including absorption to Tn, vibrational relaxation and
emission of a photon, which is now termed phosphorescence. Due to the spin
forbidden transition, this takes place on a much longer timescale, from microseconds
up to many seconds.

Triplet states play an important role in photophysical behaviour of organic
chromophores and can be manipulated, either by increasing the intersystem crossing
rate kISC using the heavy atom effect [Kasha, 1952], or depopulating triplets by using
triplet quenching molecules [Widengren and Schwille, 2000]. The importance of
selectively manipulating excited states in general lies in the fact that photophysical
reactions, for example optically induced transitions of chromophores, have several
possible pathways, e.g. higher excited singlet states, triplet states or isomerised


states. To elucidate the correct mechanism, certain pathways have to be controlled

An overview of timescales of all transitions mentioned in the JABLONSKI diagram is
given in table 2-1.

    Transition                Description                   Rate               Time (s)

   S0 → S1...Sn       Absorption (Excitation)               kexc                 10-15

   Sn → S1            Internal Conversion                    kIC              10-14-10-10

   S1 → S1            Vibrational Relaxation                 kVR              10-12-10-10

   S1 → S0            Fluorescence                           kF                10-9-10-7

   S1 → T1            Intersystem Crossing                  kISC               10-10-10-8

   S1 → S0            Non-radiative       Relaxation         knq               10-7-10-6

   T1 → S0            Phosphorescence                        kP                10-3-100

   T1 → S0            Non-radiative       Relaxation        knq,T              10-3-100

Table 2-1: Overview of possible depopulation pathways of the first excited singlet state.


The emission of a fluorescence photon is a spontaneous process based on an
electronic transition of a vibrational ground state of a higher excited singlet state, to a
higher vibrational energy level of the electronic ground state of a molecule. According
to the KASHA rule, this process usually originates from the first excited state, S1, to a
vibrational excited level of the ground state, S0. Though excitation of an electron to
higher singlet states is possible if the energy of incoming light is appropriate, there is
fast relaxation from those higher excited singlet to the first excited singlet, S1. An
exception to this rule is observed in some molecules, e.g. azulen and its derivates,
which fluoresces from its S2 state [Viswath and Kasha, 1956]. The reason for the
observation of S2 → S0 fluorescence in azulen is the relatively large S2 – S1 energy


gap, which slows down the normally very rapid rate of internal conversion from S2 to
S1 by decreasing the FRANCK-CONDON factor for radiationless transitions [Turro,

Although absorption of light resulting in electronic excitation is a completely general
experimental observation, emission of light is not. Most saturated organic molecules
and polyenes do not display efficient emission. However, if the process of
fluorescence is observed to an observable extent, the photons emitted contain
information that describe interactions of a fluorophore with its environment. The
emission wavelength, i.e. the energetic component, reflects changes in the polarity of
the medium surrounding a fluorophore. Fluorescence lifetime, the kinetic component
of emitted photons, together with the quantum yield depend critically on competing
processes which reduce the average time of population of a first excited state. In this
context, resonance energy transfer or dynamic quenching processes due to
photoinduced electron transfer have to be mentioned. Finally, polarization and
anisotropy measurements, which exploit the similar timescale of rotation from a
molecule and fluorescence, allow a description of the rotational mobility of a
fluorophore and give information about the larger molecular system the fluorophore is
attached to.


A fluorescence emission spectrum is usually the mirrored image of an absorption
spectrum. The reason for this behaviour lies in similar probabilities of electronic
transitions into excited vibrational levels, i.e. absorption, and their reciprocal
transitions, that is fluorescence. This can be easily understood if recalling the fact
that electronic excitation is a very fast process that does not greatly alter nuclear
geometry. Additionally, the spacing of the vibrational levels both in the ground state
and the excited state are comparable. The resembling vibrational structures of the
ground and excited state are the origin for the similarity of both absorption and
emission spectrum. Exceptions to the mirror rule can occur if the geometry of the
electronic excited state is clearly different, or, excimer or exciplex structures are


Generally, a fluorescence emission spectrum does not show a dependency on the
excitation wavelength for most fluorescent probes. Any excitation to higher electronic
or vibrational states leads to fast relaxation to S1, described by KASHA’s rule.

Fluorescence Lifetime and Quantum Yield

Both fluorescence lifetime, the kinetic component of fluorescence, and fluorescence
quantum yield as a measure of brightness of fluorophores, are important
characteristics of fluorescent probes. These parameters depend strongly on the
chemical structure and the environment of a fluorescent probe. It is of high
importance to choose fluorophores with the appropriate lifetime and quantum yield
for a particular experiment, e.g. high quantum yield and long fluorescence lifetime for
energy transfer experiments.

The fluorescence quantum yield Φf is the ratio of photons emitted through
fluorescence to photons absorbed, and hereby represents a measure of the
efficiency of the emission process. It is described by two types of depopulation rates
of the excited state S1, which is the radiative rate kr and the nonradiative rate knr,
according to the following equation:

                                 Φf =                                              ( 2-5 )
                                        k r + k nr

Quantum yields close to unity suggest that the predominant pathway of depopulation
of the excited state is fluorescence photon emission. In other words, the depopulation
of S1 is mostly via radiative pathways. Nonradiative transitions as relaxation or
internal conversion play, in this case, a minor role.

The determination of the absolute fluorescence quantum yield for a fluorophore is
experimentally difficult to realize. A commonly applied method is “thermal blooming”,
measuring the change in refractive index of a solvent due to a temperature increase
caused by thermal relaxation of excited molecules. More often, values for
fluorescence quantum yields are determined with respect to a fluorophore with nearly
100% quantum yield, e.g. Rhodamine6G or Rhodamine 101. Basically, quantum
yields can be determined with respect to any fluorescent probe with the absolute
quantum yield known.


The average time a molecule stays in its excited state S1, which may be referred to
the kinetic information of fluorescence emission, is described by its fluorescence
lifetime τfl,

                                   τ fl =                                            ( 2-6 )
                                            k r + k nr

In contrast to the intrinsic or natural lifetime τn, which is the lifetime of the fluorophore
in the absence of any nonradiative process, its fluorescence lifetime τfl can be
observed by optical techniques like time-resolved fluorescence measurements. As
shown in the JABLONSKI-diagram in figure 2-2, fluorescence is only one of several
possible pathways that can result from the first excited state S1. As a result, the
observed fluorescence lifetime contains information about both radiative and
nonradiative processes.

The radiative depopulation process of fluorescence is fully spontaneous and is
described by first order kinetics. Similar to a radioactive decay, the temporal
distribution of fluorescence photon emission, I(t), is described by a single exponential

                                I (t ) = I 0 exp(−           )                       ( 2-7 )
                                                     τ fl

If more than one depopulation process exhibiting fluorescence is, which is the case in
heterogeneous samples or samples with more than one fluorophore, the temporal
change of fluorescence intensity is described by

                              I (t ) = ∑ α i exp(−                  )                ( 2-8 )
                                      i                  τ i , fl

By applying time-resolved techniques, it is possible to quantify the relative
contribution, α i , of the i-th component and determine its characteristic fluorescence

lifetime, τ i, fl .

As a more general view, fluorescence spectroscopy can be classified into two types
of measurements, steady-state and time-resolved. Steady-state measurements are
characterized by constant illumination and observation of a sample and are the most
common method of fluorescence experiments. Because of the nanosecond timescale
of fluorescence, most experiments are done under steady-state conditions. Time-


resolved measurements are used to obtain kinetic information about fluorescence
emission, and require complex and expensive instrumentation like a pulsed light
source, fast electronics and sensitive detection elements. Regarding the light source
used, a narrow pulse width with respect to the fluorescence decay time is required,
which explains the common use of fast light emitting diodes (LED) or even better,
laser diodes.

Fluorescence lifetime has a characteristic value for each fluorophore and strongly
depends on any condition or environmental effect that may affect any of the rate
constants involved in depopulation processes of the excited state S1. It therefore
widens the field of application of basic fluorescence spectroscopy and microscopy.
Prominent examples that should be mentioned here are energy transfer rates
determined by measuring decreased fluorescence lifetime as well as quenching of
fluorescence taking place if a suitable fluorophore is in close proximity to quenching
molecules, e.g. tryptophan or guanosine [Marmé et al., 2003].

Measurement of Fluorescence Lifetimes

In principle, there are two widely
used        methods            for        the
measurement           of     fluorescence
lifetimes, the pulse method and the
phase modulation method. In this
work, the pulse method, i.e. time-
correlated single-photon counting,
has been preferred due to its
sensitivity and ability to deal with
low photon count rates. It has been
the method of choice for time-
resolved    experiments           in    both
ensemble        and        single-molecule

The underlying principle of TCSPC
can    be   described        as      periodical      Figure 2-3: TCSPC measurement principle.


detection of photons referred to a pulse signal. Especially in experiments with low
photon detection probability, i.e. a probability much smaller than one of detecting a
photon during one pulse cycle, several photons arriving in one period can be
neglected, and the principle shown in figure 2-3 can be used. There are many signal
periods without photons, other signal periods contain one photon pulse. Periods with
more than one photon are very rare. When a photon is detected, the time of the
corresponding detector pulse is measured. The events are collected in memory by
adding a ‘1’ in a memory location with an address proportional to the detection time,
representing a time bin. After many photon detection events, the histogram of the
arrival times representing the fluorescence decay is obtained. The main advantage of
this method lies in the fact that the accuracy of the time measurement is not limited
by the width of the detector pulse. Thus, the time resolution is much better then with
the same detector used in front of an oscilloscope or another linear signal acquisition
device. Furthermore, all detected photons contribute to the result of the
measurement. To prevent any detection signal from a previous excitation pulse in a
measurement period, the distance between two succeeding pulses is usually chosen
to be around five times the fluorescence lifetime of a measured decay. Furthermore,
the photon detection probability should not exceed a value of around 5%, to prevent
“pile-up” effects shortening the fluorescence lifetime decay. Pile-up effects occur if
occasionally, more than one photon arrives in one detection period, and since only
the first photon contributes to the histogram, photons are piled up at shorter times,
leading to the appearance of shorter fluorescence lifetimes.

In many fluorescence experiments involving laser as excitation light source, the laser
pulse width itself is of a comparable order of magnitude as the fluorescence lifetime.
Especially excitation pulses from semiconductor lasers show a full width half
maximum (FWHM) in a range of a few hundred picoseconds. As a consequence, the
observed fluorescence decay R(t) obtained by the method described is represented
as a convolution of the excitation pulse L(t) with the impulse response of a sample
that would be obtained by applying an infinitesimal small δ pulse, F(t),

                      R(t ) = L(t ) ⊗ F (t ) = ∫ L(τ ) F (t − τ )dτ            ( 2-9 )

The inverse process of deconvolution is mathematically difficult, but many different
approaches to circumvent this time consuming algorithm have been developed. Two


prominent methods are used in most experiments, allowing the extraction of exact
fluorescence decay information from the observed signal.

For most measurements, the simple least square (LS) approach is sufficient. This
method is based on finding a set of fluorescence decays, αIexp(-t/τI) , and comparing
the calculated signal, Rc(t), which is obtained after convolution with the excitation
pulse, with the measured signal, R(t). The parameters αI and τi are varied iteratively,
and the quality of the result is estimated using χ2 ,
                               χ 2 = ∑ ω i ( R(t ) − Rc (t )) 2                     ( 2-10 )
                                       i =1

where ωI is a statistical weighting factor for individual errors in each value of R(t).

If photon counts are low, as is the case in single-molecule measurements, an
alternative method appears to be more suitable, called the maximum likelihood
estimator (MLE). In particular for monoexponential decays, which is normally the
case if a single molecule is observed and no competing interactions on the same
timescale occur, the MLE offers a fast method to obtain reliable data with low error
component [Enderlein et al., 1997]. The underlying function is described as
                    1 + (e T / τ − 1) −1 − m(e mT / τ − 1) −1 = N −1 ∑ iN i         ( 2-11 )
                                                                   i =1

where T is the width of each channel, m the number of utilized time channels, N the
number of photon counts taken into account, and Ni the number of photon counts in
time channel i. The left-hand side of equation 2-11 is not dependent upon the data
and is a function only of τ, while the right-hand side is determined from the
experimental data. The lifetime can be abstracted from the data by the use of an
reiterative technique such as NEWTON’s algorithm which was applied in this work to
determine the fluorescence lifetimes from single molecules.

Fluorescence Anisotropy

If fluorescent molecules are excited by polarized light, preferentially molecules with
an absorption dipole aligned parallel to the electric field vector E will absorb the
incoming light. As a result, selective excitation of fluorophores leads to partially
polarized fluorescence emission. The transition moments for both absorption and


emission have fixed orientations within a fluorophore, and the relative angle between
these moments determines the maximum measured anisotropy. The fluorescence
anisotropy r and polarization P are defined by

                                        I vv − I vh
                                  r=                                                     ( 2-12 )
                                       I vv + 2 I vh

                                        I vv − I vh
                                  P=                                                     ( 2-13 )
                                        I vv + I vh

where Ivv and Ivh are fluorescence intensities after vertical (v) excitation measured in
the vertical and the horizontal (h) emission polarization.

If observing fluorophores freely diffusing in an isotropic solution of low viscosity, the
anisotropy will mostly be close to zero. The reason for this is rotational diffusion of
the molecules. With a typical rotation time around 100 ps, the orientation of the
fluorophores in the excited state is then randomised, and fluorescence emission does
not show any polarized component.

If the rotation time is larger than the time spent in an excited state, as is often the
case for fluorophores conjugated to large biomolecules with high molecular masses,
anisotropy measurements provide information about size and shape of these
biomolecules. Furthermore, fluorescence anisotropy can be applied to study protein-
protein interactions or interactions between proteins and nucleic acids [Lakowicz,
1999]. As a solution-based methodology, it offers a true equilibrium measure,
allowing to evaluate changes in solution conditions as salt concentration, pH, and
temperature [LeTilly and Royer, 1993].

Combining polarized excitation and detection with a pulsed light source, this method
can be expanded to time-resolved anisotropy measurements. A time-resolved
anisotropy decay is obtained and can be approximated by an exponential function,
yielding the rotation time of a fluorescent molecule. Since rotation times of small
molecules are in the order of 100 ps, a narrow excitation pulse and deconvolution
methods    are   even    more    important             than   for   time-resolved   fluorescence



If two molecules are in close proximity, interactions of different nature depopulating
the excited state without photon emission of the donor molecule may occur. These
processes, varying in their distance dependence characteristics and sensitivity, are of
strong interest for energy transfer studies, especially for the design and
characterization of multichromophoric systems.

These molecular interactions which influence fluorescence properties of molecules
can be used for the temporal observation of intermolecular distance changes below
10 nm. Any spatial or conformational change of intermolecular distance between two
interacting molecules over a large timescale from nanoseconds to seconds is
accessible, and these mechanisms are exploited in many fluorescence spectroscopic
applications [Neuweiler and Sauer, 2004].

Molecular interactions that cause quenching of fluorescence can be divided into four
basic principal mechanism, listed in table 2-2.

          Energy Transfer                      FD + FA
                                                              →         FD + FA

          Electron Transfer                    F* +Q          →       F + /− + Q−/+

          Proton Transfer                     F * + QH        →        FH + + Q −

          Exciplex/Excimer-Formation           F* + M         →          (FM )*

Table 2-2: Overview of quenching processes depopulating the first excited state of an excited
molecule F*, relevant in fluorescence spectroscopy.

Energy transfer processes can be divided into two different mechanisms. On the one
hand, there is weak coupling and non-coherent interaction, which is the case in
fluorescence resonance energy transfer (FRET), and takes place in a range of 2 to
10 nm interchromophoric distance. The mechanism of FRET is based on Coulombic
interactions of two or more chromophores, which has been theoretically derived from
classical electrodynamics [Förster, 1948] and the model of dipole-dipole coupling.
The second mechanism, i.e. electron exchange energy transfer (EEET), requires


closer proximity of the molecules in order to allow orbital interactions and is
characterized by electron exchange from a donor to an acceptor molecule.

Fluorescence resonance energy transfer (FRET) plays a major role in many of the
chromophoric systems investigated in this work. Single molecular photonic wires of
different nature have been constructed using DNA as a rigid backbone molecule,
which opens the possibility to place numerous fluorophores in well defined positions
relative to each other and thus enables resonant dipole-dipole coupling between the
chromophores. Furthermore, the principle of FRET can be used to probe
nonfluorescent dark states of acceptor molecules, that still can be in resonance with
a donor molecule. This allows for photophysical and mechanistic studies of
fluorophores, and was used to investigate the photoinduced conversion of
carbocyanine dyes.

Electron transfer processes, often referred to as photoinduced electron transfer
(PET), are part of the complex mechanism of photosynthesis, by converting light
energy into chemical energy. This mechanism involves an interaction of a
fluorophore with electron donating or accepting molecules, and results in quenching
of fluorescence through the generation of radical states. In contrast to FRET, a
collision between molecules is required. Though, electron transfer occurs only on
shorter distances, and the method allows the monitoring of smaller spatial changes,
which has been successfully applied in folding studies of peptides, proteins, DNA or
RNA biomolecules [Neuweiler and Sauer, 2004].

Photoinduced electron transfer, which results in the formation of either radical cation
or anion of a chromophore and hereby substantially changes its fluorescence
properties, is a redox active process which has been exploited in studies of
chromophores of different nature [Speiser, 1996]. By changing the redox properties
of the surrounding microenvironment of a molecule and choosing the appropriate
redox partner molecule, a longer observation time and stable fluorescence emission
of single chromophores could be achieved. Additionally, the process of switching
cyanine dyes in aqueous solution requires the presence of electron donating
molecules    in   millimolar   concentration,    which   strengthens   the   mechanistic
interpretation that radical states of the fluorophore are involved.

The reversible protonation or deprotonation of a fluorescent probe can also cause
quenching of fluorescence. The protonation of the chromophoric centre of a protein


was recently shown to be the origin of the switching behaviour of fluorescence
observed in the fusion proteins green fluorescent protein (GFP) and yellow
fluorescent protein (YFP) [Kennis et. al, 2004; McAnaney et al., 2005].

Due to their importance in this work, both FRET and PET merit a more detailed

Fluorescence Resonance Energy Transfer (FRET)

Energy transfer interactions based on
through space dipole-dipole coupling,
often referred to as weak coupling or
non-coherent coupling, are the heart
of FÖRSTER’s theory of fluorescence
resonance energy transfer (FRET)
[Förster,      1948].         In     a   simple
                                                   Figure 2-4: Schematic representation of the
mechanistic view, an initially excited
                                                   Förster transfer mechanism, FRET.
donor molecule transfers its excited
energy radiationless to an acceptor molecule via the electrodynamic coupling of both
molecules. This process, which in the case of commonly used organic chromophores
occurs at distances between 2 and around 10 nm, is depicted in figure 2-4. A large
number of reviews from many fields of different application of the FÖRSTER
mechanism have already been
published [Clegg, 1992; Clegg,
1995; Yang and Millar, 1997;
Selvin, 2000; Jares-Erijman and
Jovin, 2003], showing the impact
of resonant energy transfer.

The donor molecule in the weak
coupling limit usually emits at a
shorter     wavelength,            and   the
emission spectrum of the donor
                                                Figure 2-5: Distance dependency of FRET. The blue
overlaps     with       the        absorption
                                                line emphasizes 50% transfer efficiency for a FRET
spectrum        of      the         acceptor    pair of Cy3 and Cy5 (Förster radius, R0 = 5.6 nm).


molecule. In his work, FÖRSTER could derive a theoretical model for resonance
energy transfer, and could show that the rate of energy transfer, kT, depends on the
inverse 6th power of the interchromophoric distance, r, (see figure 2-5), via
                                                1  R0 
                                    k T (r ) =                                ( 2-14 )
                                               τD  r 

R0 is a characteristic value for a given set of chromophores, often called FÖRSTER
radius, and can be determined from the following equation,
                            9000(ln 10)κ 2QD
                     R0 =                    ∫ FD (λ )ε A (λ )λ dλ
                      6                                        4
                                                                                ( 2-15 )
                              128π Nn5  4

Here, QD is the quantum yield of the donor molecule, n is the refractive index, N is
AVOGADRO’s constant, and the integral expression describes the spectral overlap of
the fluorescence of a donor molecule, FD, and the extinction of an acceptor molecule,
εA, with respect to the wavelength, λ. This spectral overlap integral has a crucial
impact on transfer efficiency, explained by energetic overlap of the donor emission
spectrum and the acceptor absorption spectrum.

Of the experimental factors necessary for an energy transfer distance estimate, the
hardest to determine is κ2, the “orientation” factor:

                            κ 2 = (cos θ T − 3 cos θ D cos θ A ) 2              ( 2-16 )

Depending on the orientation of dipoles to each other and their mobility on the
timescale of energy transfer, this value may adapt values from 0 to 4 (figure 2-6).

In general, the actual value of κ2 is not experimentally measurable. The simplest
approximation is to assume that both donor and acceptor transition dipoles are
undergoing motion that randomises orientations much faster than the donor is
decaying to its ground state. The randomisation must be due to each probe sampling
all orientations, not due to static distribution probes. Assuming oriental randomisation
of both fluorophores, the dynamically averaged isotropic limit holds, and unhindered
and independent rotation of both dipoles yields a value of κ2 = 2/3 [Dale et al., 1979;
Torgerson and Morales, 1984]. Methods to verify the assumption of a freely rotating
fluorophore include anisotropy measurements or modulated excitation.


        Figure 2-6: Dependence of the orientation factor κ2 on the directions of the
        emission dipole of the donor and the absorption dipole of the acceptor.

Typical R0 values for a number of fluorophores that were also used to build photonic
wire molecules in the experimental section of this work, together with their quantum
yields and spectral properties, are listed in table 2-3.

                                                     ε/l mol-1 cm-1
                  λabs/nm         λem/nm                              Φf           R0/Å
                                                        (x 10 )

 RhodGreen          508             534                  0.74         0.9
    TMR             560             582                  0.95         0.9
  ATTO590           603             625                  1.20         0.8
    LCR             622             638                  1.20         0.8
  ATTO680           689             703                  1.25         0.3

Table 2-3: Spectroscopic parameters and R0-values of some chromophores used for the
construction of DNA-based photonic wires.

The efficiency of energy transfer, E, which is the fraction of photons absorbed by the
donor and transferred to the acceptor, is given by

                                   E=                                                  ( 2-17 )
                                        τ + kT


As a result, the efficiency E is the ratio of the transfer rate to the total decay rate of
the donor molecule, and is easily rearranged using equation 2-14 to
                                  E=                                              ( 2-18 )
                                         r 6 + R0

In practice, fluorescence energy transfer efficiency is typically determined in two
different ways [Clegg 1992]. One possible method is to determine the relative
fluorescence intensity of the donor both in the absence and presence of the acceptor,
FD and FAD. Similarly, fluorescence lifetime values under the same conditions, τD and
τΑD respectively, can be used to determine E:

                                   E = 1−                                         ( 2-19 )

                                               τ DA
                                      E = 1−                                      ( 2-20 )

Both equations 2-19 and 2-20 are only applicable for a pair of chromophores with
fixed distance and a homogeneous sample. If fluorescence lifetimes are used to
determine E and no single exponential decay is observed, it is important to use
average lifetimes, given by the sum of the αiτi products, where αI describes the
relative fraction of a component i.

The distance dependency of FRET and the range of 2 – 10 nm typical for this type of
energy transfer suggests its use as a spectroscopic ruler [Stryer and Haugland,
1984; Dietrich et al., 2002]. In most experiments, it is more common to use relative
distance changes, hereby selecting the range from 0.5R0 to 1.5R0. In this range, the
important slope of E with respect to the distance can be exploited. As an example for
larger distances: assuming a distance of r=2R0 yields to a relative energy transfer
rate decreased to 1.56%. Nevertheless, using short and rigid molecular scale
molecules like DNA or polyprolines, FRET is sometimes used to determine absolute
distances and subpopulations [Schuler et al., 2005].


Electron Exchange Energy Transfer

In the case of electron exchange
energy      transfer,     much     shorter
distances    between      molecules      are
needed. The mechanism involves an
electron exchange step between the
lowest unoccupied molecular orbital
(LUMO) of an acceptor A and the
semi occupied           molecular orbital          Figure      2-7:   Schematic    representation   of
                                                   electron exchange mechanism.
(SOMO) of a donor D, depicted in
figure 2-7. Theoretical work was first carried out by DEXTER [Dexter, 1953] who
proposed an exponential distance dependency for the rate of electron exchange
energy transfer, kEEET,

                                 k EEET = KJe ( −2 RDA / L )                                   ( 2-21 )

In the proposed model, kEEET depends on the spectral overlap integral, J, of donor
and acceptor, which is normalized to the extinction coefficient of the acceptor A and
thereby independent on the acceptor’s absorption characteristics, which is in contrast
to the mechanistic model presented for FRET (see equation 2-15). Further, K
represents a specific constant for a specific orbital interaction, RDA is the distance
between donor and acceptor and L is the sum of the                    VAN DER     WAALS radii of both

Photoinduced Electron Transfer (PET)

Light-induced transfer of an electron occurs due to changes in the redox properties of
excited molecules. In those cases, molecules are more potent electron donors or
acceptors. If a quencher molecule is present a redox reaction will take place. Like
any electrochemical reaction based on electron transfer, this is only possible if free
energy is won, which can be estimated from the electrochemical potentials of the
participating molecules.


 Figure 2-8: Simplified schematic presentation of photoinduced electron transfer, left:
 reduction of fluorescent molecule in its excited state F* by an electron donor D, right:
 oxidation by an electron acceptor A. kCS and kCR denote the rate of charge separation and
 recombination, respectively.

After excitation of a donor molecule by light, one electron from the highest occupied
molecular orbital (HOMO) is transferred to an energetically higher semi occupied
molecular orbital (SOMO). Depending on the redox potential of the excited molecule
and a second molecule around, two reaction pathways are possible. In the first,
which represents the oxidation of the first molecule, the electron from the SOMO is
transferred into the lowest unoccupied molecular orbital (LUMO) of an acceptor
molecule, A, forming a radical cation. In the reduction path, an electron from the
HOMO of a donor molecule, D, is transferred into the lower SOMO of the excited
fluorescent molecule, generating a radical anion and inhibiting fluorescence (figure 2-

Whether an electron transfer is feasible depends on the change of the free energy of
the charge separation process, ∆GCS , which can be estimated by the oxidation and

reduction potentials, Eox and Ered, the transition energy E0,0 and the COULOMB
potential of the charge separated state, ∆GCoul ,

                         ∆GCS = E ox − E red − E 0,0 + ∆GCoul
                           0                             0
                                                                                    ( 2-22 )

If fluorescence quenching experiments are carried out in the same solvent used for
the determination of the redox potentials, solvation energies of the radical ion pair


can be neglected and the free reaction enthalpy can be calculated using the classical
REHM-WELLER equation [Rehm 1969]:

                                                − e2
                                    ∆GCoul =
                                                                                           ( 2-23 )
                                               ε S RC

This equation describes the distance dependency of the coulombic term RC in PET to
be ~1/RC, including the solvent dielectric constant, εS.


Molecular oxygen is an important participant in photochemical processes because of
its high chemical energy content, its unique reactivity characteristics and its low lying
excited states. In many reaction systems, the ubiquitous presence of molecular
oxygen    influences   the   photophysics           of    fluorophores.     At   single   molecule
concentrations, the concentration of oxygen becomes important as the probability of
a collision with a fluorophore increases. The result of such a collision depends on the
energy of photophysical states involved in the reaction, and subsequent reactions,
e.g. photobleaching due to epoxide formation, are possible.

Molecular Description

In its ground state, oxygen appears as a biradical, and can be described by the
following electron configuration:

                             O2 → (core)(π x ) 2 (π y ) 2 (π x )1 (π * )1

where π x and π y denote binding π-orbitals filled each with two paired electrons, and

π x and π * are unbinding π-orbitals filled each with one electron that have parallel

spin orientation to each other, presenting a triplet ground state. Considering the
orbital occupancies of the unbinding π-orbitals filled with two electrons, four possible
occupancies are possible. Besides the triplet ground state, three possible excited
states with singlet character can be constructed. The electron orbital occupancy
corresponding to all four states, together with energy levels and spectroscopic
annotations, are shown in figure 2-9.


    Figure 2-9: A qualitative description of the three lowest electronic orbital configurations and
    states of molecular oxygen.

The pure radiative lifetimes of the singlet states 1∆ and 1Σ are relatively long, 2.7x103
s and 7.1 s, respectively [Badger et al.,1965], but are usually not observed under
laboratory conditions because of efficient deactivation due to chemical or collisional
quenching. In solution, the lifetime of                 ∆ depends strongly on the chemical
surroundings, and longest values of around 1 µs are obtained in solvents considered
to be chemically inert. The lifetime in water is around 2 µs, and around 30 µs are
observed in acetonitrile [Merkel and Kearns, 1972]. The lifetime of the higher
energetic 1Σ is not known and assumed to be too short-lived for any further reaction.

Generation of singlet state oxygen 1O2

In its triplet ground state, molecular oxygen can react with a photosensitizer
molecule, which is commonly a strong absorbing dye, producing singlet oxygen in the
    ∆ state, following the scheme:

                                  Dye + hυ →1Dye→ 3Dye                                      ( 2-24 )

                                       Dye+ 3O2 → Dye+1O2                                   ( 2-25 )

The mechanism underlying equation 2-25 represents a triplet-triplet annihilation
which results in an energy transfer to yield singlet oxygen. It is generally assumed
that an electron exchange energy transfer mechanism operates in the triplet-
photosensitized formation of singlet oxygen [Turro, 1991]. Based on the theoretical
work of DEXTER [Dexter, 1953], electron exchange energy transfer is characterized by


a collision of both molecules, and hereby enabling direct interaction of the orbitals
involved. The process of triplet-triplet annihilation and energy transfer following the
FÖRSTER mechanism is not allowed, since the spin orientation of the acceptor
molecule is necessarily changed [Förster, 1948].

Photochemical mechanisms for generating singlet oxygen are radical-like in nature,
since the key interactions involve diradicaloid structures. The thermal generation of
singlet oxygen usually involves two electron (zwitterionic) processes. A synthetically
used path to generate singlet oxygen is the decomposition of ozonides, e.g.
phosphite ozonides or endoperoxides [Turro, 1991].

Quenching of Excited Singlet and Triplet States by Molecular Oxygen

Ground state molecular oxygen is a general and efficient quencher of the S1 and T1
states of organic molecules. The mechanism of quenching can be either physical or
chemical in nature. The most common chemical quenching mechanisms are
diradicaloid electron transfer and addition:

                                  & &           &−
                            M * + O − O → M + + O2                             ( 2-26 )

                                 & &    &     &
                            M * +O −O → M −O −O                                ( 2-27 )

Physical quenching mechanisms include exciplex formation and energy transfer. In
general, the quenching of S1 of aromatic hydrocarbons occurs close to the diffusional
rate. Quenching of T1 occurs within an order of magnitude of the diffusional rate, but
is consistently slower.

An organic fluorophore therefore can interact with molecular ground state oxygen
from its S1 or T1 state, depending on the energy of the SOMO in its photoactivated
state. By removing oxygen, it is possible to find out the more important reaction
pathway. Certainly, this depends on the chemical structure of a fluorophore and has
to be explored for each molecule separately.



Whether or not efficient emission of a molecule is observed is determined by the
radiative rate constant kr, the fluorescence quantum yield Φ f and the excited state

lifetime τn. These parameters are closely related to the chemical structure of a
molecule on the one hand, and sensitive to experimental conditions on the other
hand. As an overview, a number of different classes of fluorophores used in
fluorescence microscopy are presented in the following section.

Organic Fluorophores

The group of synthetically designed
organic fluorophores is the most
widely    used    in      fluorescence
spectroscopy. A large number of
fluorophores with wavelengths from
around 350 nm to 800 nm are
commercially available, and may be
modified chemically to allow further
conjugation   reactions     to    other
molecules of interest, e.g. proteins,
nucleic acids and others.

Organic chromophores exist in many
basic structures. Most commonly
used and known are cyanines (e.g.
Cy5,     Amersham        Biosciences),
rhodamines (e.g. Rhodamine Green)
and oxazines (e.g. MR121, AttoTec-
GmbH),    representing      the   main
                                          Figure 2-10: Structure of three dyes representing
classes of dyes used in this work.        main classes of chromophores: Carbocyanine Cy5,
The chemical structures of one            rhodamine   derivative   Rhodamine   Green   and
chromophore of each type is shown         oxazine dye MR121.


in figure 2-10. Since they are all chemically distinct and show largely different
photophysical properties, an important part of this work was dedicated to intensive
studies of organic chromophores under different conditions, e.g. pH-value, redox
potential and solvent.

Absorption and emission of an organic chromophore can easily be described by
using a simple quantum mechanical approach of a particle-in-a-box. In this principle,
only the possible path of electron delocalisation is taken into account. This example
shall be briefly discussed with respect to cyanine dyes, Cy3 and Cy5, illustrated in
figure 2-11. In each of these dyes the polymethine chain forms a conjugated chain
extending from the nitrogen atom on one end of the molecule to the nitrogen atom on

 Figure 2-11: The principle of a particle-in-a-box applied to cyanine dyes, Cy3 (left) and Cy5
 (right). The polymethine chain (green) connecting both indole moieties represents the box
 where π electrons (blue) are delocalised.

the opposite side of the molecule. These nitrogen atoms are assumed to be the walls
of a one-dimensional box of length L. Assuming that the electrons do not interact, the
potential energy along the chain is essentially zero and sharply rises to infinity at the
ends of the chain. With this approximation, the length of the box is kb, where k is the
number of bonds along the polymethine chain and b is 139 pm, the carbon-carbon
bond length in benzene. The energy levels of this particle-in-a-box system are given

                                    En =                                               ( 2-28 )
                                             8me L2

where n is the quantum number, me is the electron mass, and h is PLANCK’s constant.
The number of π electrons and the length of the polymethine chain in each dye are
related to the number of double bonds between the nitrogen atoms. Each carbon


atom in the chain donates one π electron and the two nitrogen atoms donate a total
of 3 electrons to form a mobile cloud of electrons along the conjugated chain (above
and below the plane of the chain). The PAULI exclusion principle allows no more than
two electrons in each energy level. For molecules with an even number of π
electrons, N, the ground state will have N/2 filled energy levels. Electronic transitions
can occur from filled to unfilled levels when light of the appropriate energy is
absorbed. The lowest transition energy will be that of an electron jumping from the
highest occupied molecular orbital (HOMO), where nHOMO = N/2, to the lowest
unoccupied molecular orbital (LUMO) with nLUMO = (N/2 + 1). This is the transition
observed in absorption spectra of the dyes. The energy change upon excitation of
one electron is given by:

                           h 2 (n LUMO − n HOMO ) h 2 ( N + 1)
                      ∆E =                       =                                ( 2-29 )
                                   8me L2           8me L2

The wavelength of the transition is

                                        8me cL2
                                   λ=                                             ( 2-30 )
                                        h( N + 1)

where c is the speed of light.

Applying these theoretical assumptions to the polymethine dyes Cy3 and Cy5 in
figure 2-11, we obtain an approximation of the absorption wavelength of Cy3 and
Cy5, summarized in table 2-4:

                                                            Cy3      Cy5

               Number of π electrons involved                8        10

               Box length L                               1.11 nm   1.39 nm

               Calculated absorption wavelength           453 nm    579 nm

               Measured absorption wavelength             550 nm    649 nm

Table 2-4: Comparison of calculated and measured absorption wavelengths of Cy3 and Cy5.

Although a difference of 70 nm and 97 nm in absolute values is obtained, this
calculation allows a rough estimation of absorption wavelengths and is a good
example to understand the basics of chromophore structure. It has to be mentioned


that a modification of the box length L was made according to the assumption of
KUHN who suggested adding one extra bond length at each side of the polymethine
chain [Kuhn 1959]. This procedure balances out the non-vertical potential jump
occurring at the nitrogen atom site. If we assume a perfect box and no interaction
between electrons, the expected box length for Cy5 and Cy3 using this simple model
would be 1.39 nm and 1.11 nm, respectively. Extended to Cy7, an infrared cyanine
dye with similar structure but one additional methine bridge, the difference of
calculated and experimentally derived values decreases. The calculated absorption
maximum is at 706 nm, whereas 743 nm are measured. This shows the critical
influence of the box length, which is the weak point of this simple model.

In a more general view, organic
chromophores         show          the
expected broad absorption and
fluorescence spectra as expected
from theoretical considerations,
involving    mixed   transitions    of
electronic and vibrational type
(often termed vibronic). Figure 2-
12 shows the absorption and
emission      spectrum      of      a
previously      presented          and
                                              Figure 2-12: Absorption and fluorescence emission
commonly used fluorophore, the                spectrum   of   the   organic   chromophore     Cy5,
cyanine dye Cy5.                              conjugated to a short 20bp DNA strand.

Whether a fluorophore is suitable
for fluorescence microscopy experiments depends on its photostability. This is, in
other words, a measure for the number of excitation cycles a molecule undergoes,
before it is irreversibly destroyed, i.e. photobleached. This results in a limited number
of photons emitted by a fluorophore. Photobleaching of a dye can be characterized
by the quantum yield of photobleaching Φ PB , which is defined as

                                              N PB
                                     Φ PB =                                            ( 2-31 )
                                              N abs

where Nabs is the number of photons absorbed, and NPB is the number of molecules
photobleached [Eggeling et al., 1998]. Prominent examples for poor photostability are


coumarin dyes and carbocyanines, whereas rhodamines and oxazines are more
photostable. One possible pathway for photobleaching is the interaction between
molecular oxygen and a fluorophore [Christ et al., 2001] (discussed in detail in
section 2.1.3). Other pathways of photobleaching originate from higher excited
singlet or triplet states. An increasing antibinding character of these states leads to
destabilized and more reactive molecules and favours photodestructive reactions.

One focal point in this work was dedicated to investigate the influence of a change in
the microscopic environment of a fluorophore on its photostability. Here, a change in
pH, redox potential, solvent or oxygen concentration has to be mentioned. Finally,
the aim was to find specific stabilizing conditions for different types of fluorophores
which then allow to build complex multichromophoric systems, e.g. photonic wires,
and make single-molecule measurements possible. The accuracy of the observation
of single multichromophoric systems depends critically on the number of detected
photons, so stabilized and “heterogenized” systems are desirable. Longer
observation times can be achieved, and a better accuracy of interpretations can be

Besides being influenced directly via the local environment, a single fluorophore can
sometimes be manipulated externally by physical means, e.g. light. As an example,
an interesting result obtained during this work is the laser-induced reactivation of a
fluorophore in a temporal dark-state by a second wavelength of excitation light,
performed with cyanine dyes, namely Cy5 and Alexa647 (Molecular Probes, USA).
Attached to DNA and exposed to a reductive environment, these molecules show a
reversible switching behaviour between an “on”- and an “off”-state which can be
generated using two different laser wavelengths, one near the absorption maximum
of the chromophore, and a second with 150 nm shorter wavelength.


Quantum Dots

A new class of fluorescent probes which have in recent years become commercially
available are colloidal quantum dots, which are nanometer-sized crystalline
structures of semiconductor materials (figure 2-13). Today, quantum dots spanning
the whole visible spectrum are available, the surface can be functionalised with
biomolecules or chemical groups activated for conjugation reactions, and they are
used in many different areas of research today [Alivisatos, 2004; Michalet et al.,

If a semiconductor quantum dot, e.g. with
cadmium selenide (CdSe) as core material, is
excited    with   light,   charge    is   separated
compared to the ground state, and an excitonic
state or “exciton” is created. This process
results in the transfer of an electron being
located mostly on selenium atoms to cadmium
atoms, which corresponds to a transition of an
                                                        Figure 2-13: Structure of a Quantum
electron   from    the     valence   band   to    the
conduction band of the semiconductor. By
doing this, the electron leaves behind a positive charge or a hole, and creates a
negative charge in a new location. Fluorescence emission of quantum dots is
described as the recombination of this electron-hole pair, or exciton. The wavelength
of light emission depends solely on the size of the quantum dot, following the model
of a particle-in-a-box previously described for organic fluorophores. Therefore, the
smaller the radius, the shorter fluorescence emission occurs.

Usually, fluorescence lifetimes of quantum dots are in the range of 10 to 100 ns. As
another interesting property, a unique broad band excitation spectrum allows
simultaneous excitation of different quantum dots emitting at different wavelengths,
which has been exploited for colocalization studies [Lacoste et al., 2000; Michalet et
al., 2001]. Furthermore, photobleaching is a rare event, and so they allow collection
of a much higher number of photons than any other fluorescent probe. Nowadays,
the use of quantum dots in biology has been demonstrated in-vivo [Smith et al.,


2004], but the applications are still limited, which is mainly due to deficiencies in
conjugation chemistry and to the size.

Metallic Nanoparticles

Besides       quantum           dots,        which
represent     a     huge     class      of    well
described and confined nanocrystal
structures     exhibiting       fluorescence,
many metallic structures have also
been       found    to     be     fluorescent.
Prominent examples here are size-
tunable AgnO-clusters or Aun-clusters
[Zheng et al., 2003]. Silver oxide
clusters      are     less        suited       for         Figure 2-14: Fluorescence excitation and
conjugation        chemistry      than        gold         emission spectrum of Au8 clusters.

clusters that can easily be conjugated
to sulphydryl groups, but the use of either of these in a similar way like organic
chromophores or quantum dots is currently not in sight.

Fluorescence spectra of recently discovered and brightly fluorescent gold dots Au8
are shown in figure 2-14. Synthesized in a polymer matrix, they exhibit an absorption
maximum at 375 nm and a fluorescence maximum around 460 nm. A
monoexponential fluorescence lifetime of 7.4 ns was determined.


Fusion Proteins

Another class of fluorescent molecules, derived
directly from nature, is represented by a number
of fusion proteins, containing a chromophoric unit
embedded into a protein’s tertiary structure. The
first   fluorescent   protein   which   had      been
discovered was the green fluorescent protein
(GFP) from aequorea jellyfish [Shimomura et al.,
1962]. A three dimensional view of this 27 kDa
protein is depicted in figure 2-15. It can be
described as a β-barrel structure embedding the
chromophoric unit in the inner of the protein. The      Figure 2-15: Green fluorescent
chromophoric unit in the protein is spontaneously       protein (GFP) from jellyfish.

formed by three amino acids, i.e. glycine, tyrosine
and threonine (or serine), and totally shielded from the surrounding environment by a
cylindrical can of the protein backbone.

In the last decades, GFP and similar fluorescent proteins have been widely used in
fluorescence microscopy (a comprehensive review is given in [Tsien, 1998]). The
major advantage of fusion proteins is the ability to express the target protein directly
fused to the desired fluorescent protein by “melting” the genetic information of both.
In most cases, the functionality of the modified protein is hereby conserved. This
simplifies the use of fusion proteins in biological assays, especially in-vivo, since
labelling techniques or invasive treatment of organisms are not necessary.

Since the discovery of GFP, many mutants of this protein have been derived, but
also other natural fluorescent proteins from other organisms have been discovered. A
short overview of a small number of representative proteins is given in table 2-5
[Shimomura, 1979; Baird et al., 2000; Tsien, 1998].


          Protein                              Organism           λexc      λem

          Green Fluorescent Protein, GFP    Aequorea jellyfish   480 nm   509 nm

          DsRed                             Discosoma            558 nm   583 nm

          Yellow Fluorescent Protein, YFP   Mutant of GFP        514 nm   527 nm

          Cyan Fluorescent Protein, CFP     Mutant of GFP        402 nm   468 nm

         Table 2-5: Overview of a number of fusion proteins exhibiting fluorescence.

Both GFP/dsRED and YFP/CFP represent FRET pairs commonly used in biological
experiments [Chan et al., 2001; Kluge et al., 2004]. Extended to two-photon FRET
imaging microscopy, this technique can provide details of specific protein molecule
interactions in living cells [Chen and Periasamy, 2004]. Very recently, three-
chromophore FRET using the three fusion proteins CFP, YFP and RFP was used to
investigate multi-protein interactions in living cells [Galperin et al., 2004].

An interesting property of the chromophoric unit of fusion proteins which was first
discovered in 1997 for GFP and is related to one important topic in this work is on/off
blinking and switching behaviour [Dickson et al., 1997]. After preparing a dark state
of a single GFP unit by irradiation with 488 nm over 90 s, fluorescence could be
recovered by irradiation with 405 nm for about 5 min. This procedure could be
repeated several times. Similar optical switching behaviour was found for CFP
[Chudakov et al., 2004] and YFP [McAnaney et al., 2005]. The mechanism of
reversible photoswitching was found to go hand in hand with proton transfer at the
central chromophoric unit of the fusion protein [Stoner-Ma et al., 2005].

2.2. Single-Molecule Spectroscopy

First realized in 1989 by MOERNER and KADOR at low temperature [Moerner and
Kador, 1989], optical single-molecule spectroscopy (SMS) based on detection of
fluorescent photons of single fluorophores at room temperature has found many
applications [Weiss, 1999]. Nowadays, laser-induced fluorescence detection is used
for various ultrasensitive analytical techniques in chemistry, biology and medicine by
probing reagents that are either autofluorescent or tagged with a fluorescent dye
[Eigen and Rigler, 1994; Goodwin and Ambrose, 1996]. In contrast to the averaging


process inherent to ensemble measurements, the observation of single molecules
reveals subpopulations with different nanoenvironment.

In this work, single-molecule fluorescence measurements were performed on the
basis of a confocal microscope. Experiments were carried out both in solution and on
dry surfaces. State-of-the-art instrumentation and software allowing for single-photon
analysis using photon-histogram techniques and autocorrelation, together with
fluorescence lifetime and spectral information were exploited for the characterization
of single fluorophores or multichromophoric systems. Together with single-pair FRET
(spFRET) and multistep single-pair FRET (multi-spFRET), these principles shall be
discussed in the following section.


The basic principle of confocal microscopy was first established by MINSKY in the
1950s (and patented in 1957) [Minsky, 1957]. This technique offers several
advantages over conventional widefield optical microscopy, including the ability to
control depth of field, reduction of background originating above or below the focal
plane, and the capability to collect serial optical sections from thick specimens. The
key principle of the confocal approach is the use of spatial filtering techniques to
eliminate out-of-focus light.

In 1992, RIGLER and co-workers first used confocal microscopy to detect single
molecules [Rigler and Mets, 1992; Rigler et al., 1993]. By reducing the focal volume
of excitation and applying the confocal principle, background is reduced considerably
such that the technique has evolved to the most common one used in single
molecule spectroscopy in the last decade.

Resolution Limit in Microscopy

In the middle of the 19th century, it was the work of ERNST KARL ABBÉ that led to the
emergence of microscopic techniques. By developing the first achromatic microscope
objective and making microscopy available to many fields of science, he founded
modern light microscopy.

To generate an image in microscopy, a parallel light beam from an excitation source
must be diffracted by a sample. Together with an objective lens, the resolution limit of


the microscope is defined. If we assume the observation of an periodic grating with a
spacing dmin that can just be resolved, the resolution obtained is given by

                                               λ           λ
                                  d min =             =                                     ( 2-32 )
                                            n sin α        NA

where n is the refractive index of the medium, α is the opening angle of the objective,
NA is the objective’s numerical aperture, and λ is the wavelength of the diffracted

A        complementary      method       for
defining the limit of resolution was
presented by HELMHOLTZ and uses
point objects instead of line gratings.
The image of an infinitively small
luminous object point, i.e. a point
source      like   a   single   fluorescent
molecule, is itself not infinitively
small, but is a circular AIRY diffraction
image with a bright disk in the centre             Figure 2-16: Intensity distribution described by
and progressively weaker concentric                the AIRY function.
bright and dark rings. The intensity
profile of the AIRY-disk is described by
                                              J (r ) 
                                    I (r ) ∝  1                                           ( 2-33 )
                                              r 

where J1(r) is a BESSEL function of the first order and the first kind. This intensity
distribution is also called the AIRY pattern, shown in figure 2-16, and represents the
lateral intensity distribution of the point spread function (PSF) of a microscope in
optical units.

The radius of the first dark ring around the central disk of the AIRY diffraction image,
rAiry, which equals the distance of the first minimum to the principal of the intensity
function, depends on λ and NA of the objective:

                                     rAiry = 0.61                                           ( 2-34 )


If there exist two equally bright points in the same image plane, they are said to be
resolved if their distance d to each other is larger or equal to the radius of the AIRY
disk. This is called the RAYLEIGH criterion and relies on the assumption that two point
sources radiate monochromatically and incoherently (whereas the approach of ABBÉ
assumes coherent excitation). The intensity between both point sources drops of to
76% of the maximum intensity [Born and Wolf, 2002].

The axial resolution, i.e. the resolution along the optical axis of the microscope or z-
axis, is defined using the three-dimensional diffraction image of a point source that is
formed near the focal plane. The distance from the centre of a three-dimensional
diffraction pattern to the first axial minimum is given by

                                  z min =                                       ( 2-35 )
                                            ( NA) 2

where n is the refractive index of the object medium. zmin corresponds to the distance
a mircroscope objective has to be raised in order to focus the first intensity minimum
observed along the microscope axis of a three-dimensional intensity pattern instead
of the central maximum. Comparable to the lateral resolution image, zmin is a
measure of the axial resolution of the microscope optics. In contrast to the lateral
resolution, the axial resolution decreases with the square of the numerical aperture of
the objective, NA.

These considerations for resolution assume that the object is viewed in conventional
wide-field (WF) microscopy. In the case of confocal microscopy, where the field of
view becomes extremely small, the resolution can in fact be greater than in wide-field
techniques. This phenomenon is the issue of the following paragraph..

The Confocal Principle

The optical path of an epi-illuminated confocal laser scanning microscope is
characterized by an identical focus used both for excitation and for detection [Pawley,
1995]. A scheme of the optical path is shown in figure 2-17. Parallel light emitted by
the laser system (excitation source) enters into the rear aperture of a microscope
objective via a dichroic mirror, which is used to separate excitation light from
fluorescence emission of the sample. The fluorescence signal (red in figure 2-17) of
the sample now passes back through the dichroic mirror and, by passage through a


second lens, is focused as a second
focal point onto the detector pinhole
aperture.   Finally,     the    signal    is
recorded by a detecting device, e.g. a
photomultiplier    or     an    avalanche
photodiode. Not that if light does not
come from the focal plane of detection
(blue lines in figure 2-17), it will not
pass the pinhole aperture and though
not reach the detector.

As a result, the field of view in
confocal      microscopy         becomes
extremely small, and only fluorescent
and light scattering objects in the axial
and lateral resolution are viewed.
Compared to the classical resolution              Figure 2-17: Optical path of an epi-illuminated
                                                  confocal   microscope.    The   excitation   light
limit   achieved    by     applying      the
                                                  (green) is reflected by a dichroic mirror,
RAYLEIGH criterion (or more precisely,
                                                  whereas fluorescence (red) is let through. Out
by applying the criterion of 76% sum              of focus light (blue) does not reach the detector.
intensity   between       two    point-like
objects), an increase for both lateral and axial resolution by a factor of 1.41 is
achieved in confocal microscopy,

                                  rlateral = 0.40                                           ( 2-36 )

                                  raxial = 1.41                                             ( 2-37 )
                                                  NA 2

The factor of 1.41 is derived from the square of the AIRY intensity distributions (see
eq. 2-33) in confocal microscopy caused by a focussed excitation light source.

In this work, a high numerical aperture oil immersion objective with NA = 1.45 (100x,
Zeiss, Jena, Germany) was used for most experiments. A calculation of resolution
considering as example two common emission wavelengths of 530 nm and 670 nm
(as used in this work), one obtains a maximum lateral resolution of about 150 nm and


185 nm, and an axial resolution of about            525 nm and 665 nm (immersion oil
refractive index n = 1.52 at room temperature).

The major advantage of this technique over non-confocal detection is that only light
coming directly from the focal plane will be detected, light above and below this plane
does not reach the detector. This allows confocal scanning microscopy by scanning
an object in all three spatial dimensions, x,y and z, and hereby creating a true three
dimensional image. Due to the optical sectioning realized by applying small diameter
laser foci, stray light is minimized, resulting in clearer images with higher resolution.

Pinhole Aperture

The primary reason a pinhole is used in the detection path of a confocal microscope
is to reduce the amount of stray light or, generally spoken, light originating from out-
of-focus objects. The size of a pinhole is hereby related to the size of the AIRY
function, representing the point spread function of a small light emitting object.

The FWHM of the AIRY function for a microscope objective with NA = 1.45 and a
magnification of M = 100, together with an emission wavelength, λ, has a size of 55
µm, calculated from the position of the first minimum,

                                        3.77 Mλ
                                   r=                                                ( 2-38 )
                                         2 NAπ

To achieve the best possible resolution, the size of the pinhole is chosen to 50 to
60% of the FWHM of the AIRY disk, which would suggest a diameter of 28 to 32 µm.
Since increasing the resolution goes hand in hand with a reduction in collected
photons, a pinhole in single-molecule research commonly has a diameter beginning
with 50 µm, and, at low signal to noise ratios, even 100 µm or larger. A second factor
to be taken into account is the fact that the pinhole position in the detection path,
situated in the image plane of the microscope, is critical and any misalignment leads
to severe loss of photons and reduced resolution, since out-of-focus light is projected
onto the detectors. Especially in experiments requiring detection of a wide range of
different wavelengths, achromatic behaviour of any component of the microscopic
system makes the resolution dependent on the wavelength of light detected.
Therefore, a 100 µm pinhole which compromises both resolution and low photon
statistics was used in most experiments presented in this work.



To achieve the optical detection of a single chromophore, the advantages of confocal
microscopy have been applied, and the microscopic system has been modified to
realize a high detection sensitivity in order to be able to collect single photons emitted
from single chromophores.

Single-Molecule Event

The detection of a single molecule can theoretically be achieved by stepwise dilution
of a solution until the concentration of molecules in a certain time bin is around one
per detection volume. Assuming a typical excitation and detection volume of a few
femtoliter in confocal microscopy, this is achieved with concentrations below 10-8 M.
Since typical diffusion times of molecules through the laser focus are in the range of
hundreds of microseconds, a probability to detect one molecule per second is
observed for concentrations of ~10-11 M.

Similar, the dilution criterion can be applied to measurements on dry surfaces, or
molecules immobilized at a surface, surrounded by a liquid environment. Generally, a
density of about 1 molecule per µm2 is well suited to most point spread functions and
allows undisturbed detection of photons originating from one single fluorophore. As
opposed to diffusion-controlled experiments in solution, the observation of single
molecule fluorescence trajectories reveals some typical effects which are commonly
attributed to the presence of a single chromophore, e.g.

              •   Fluorescence intermittency (“blinking”) due to reversible transition
                  into a dark state

              •   Irreversible photobleaching due to photo-induced reaction of the

              •   The correlation between emitted photons exhibits the signature of
                  antibunching [Harnbury-Brown and Twiss, 1956]

              •   Polarized excitation exhibits the presence of a well defined
                  absorption and emission dipole.


Most of the arguments mentioned above provide evidence for singleness.
Unfortunately, some fluorophores have very unique optical properties that can
interfere with the proposed criteria. One-step photobleaching, discrete intensity
fluctuations and spectral jumps have also been observed in multi-chromophoric
systems such as light harvesting complexes [Bopp et al., 1997; van Oijen et al.,
2000], conjugated polymers [Vanden Bout et al., 1997; Yip et al., 1998; Huser et al.,
2000] or polyphenylene dendrimers [Hofkens et al., 2000]. B-Phycoerythrin, a protein
playing an essential role in light harvesting processes of cyanobacteria, have actually
been found to be comprised of several fluorophores, which act effectively as one
quantum emitter [Wu et al., 1996]. Spherical quantum dots, for example, have two
degenerated emission dipoles [Empedocles et al., 1999] making it more complicated
to prove singleness by polarization measurements.

Typical single-molecule signatures of fluorophores immobilized onto a protein surface
under aqueous conditions are shown in figure 2-18.

   Figure 2-18: Typical single molecule fluorescence signatures, (a) immobilized at protein
   surface under aqueous conditions, (b) fluorescence trajectory recorded from one single
   fluorophore, both exhibiting blinking and bleaching of single fluorophores. (Atto647
   conjugated to a double strand 60bp DNA, anchored to surface by biotin linker, PBS).


One commonly observed source of single-molecule intensity fluctuations involves
discrete intensity jumps from an “on” (high) to an “off” (background) intensity level
This effect, which is usually denoted by the term “blinking”, is due to quantum jumps


of the single molecule to long-lived nonemissive “dark” states. A common source of
blinking in organic fluorophores involves intersystem crossing to a long-lived triplet
dark state. Other well-known sources for intensity fluctuations are cis-trans-
isomerizations [Widengren and Schwille, 2000], spectral jumps or the formation of
radical states [Yip et al., 1998].

Detailed analysis of single-molecule intensity fluctuations can provide insight into the
underlying excited-state dynamics and help in the classification of photophysical
processes. Two common methods that were used in this work are intensity duration
histograms and autocorrelation analysis of single-molecule fluorescence fluctuations.
Both methods are presented in the following.

Duration Histograms

Especially if fluorescence count rates of single molecules are high and fluctuations
between a fluorescent and a nonfluorescent state have a digital character with two

 Figure 2-19. Procedure of deriving kinetic information about a single emitter using the
 duration histogram method. Assuming a simple three-state model, (a), with a longlived
 nonfluorescent triplet state appearing in fluorescence emission, (b). “on” and “off” periods
 describe transitions between energetic levels, (c), and are summed in a histogram, (d).
 Characteristic times and rates are extracted from histograms by approximation with
 exponential functions.


well-defined intensity levels, duration histograms allow to obtain kinetic information
down to around hundreds of microseconds [Yip et al., 1998]. Briefly described, the
fluorescence emission is classified into “on” and “off” periods. The “lifetime” of these
periods is then summed up in histograms. The bin width used to histogram these “on”
and “off” states should be the lowest possible time resolution which allows to
differentiate between the emission levels unambiguously. Duration histograms
obtained with this procedure can now be approximated by exponential functions,
yielding characteristic transition times, τ on and τ off , and transfer rates kon and koff,

respectively. A summary of this procedure is presented in figure 2-19.

The kinetic parameters necessary to completely describe intensity fluctuations due to
triplet blinking within this model are τ on , the average lifetime of the kinetic “on” state,

during which the molecule is cycling between S0 and S1, τ off , the average lifetime of

the dark state, and Ion, the single molecule fluorescence intensity during the “on”
cycle. Together with the expression for the excitation rate kexc,

                                  k exc = I excσ (λ )γ                                ( 2-39 )

where Iexc is the excitation power, σ (λ ) is the absorption cross section of a molecule
and γ a correction factor, the following relationships of these quantities are obtained:

                                  τ on = (k excφisc ) −1                              ( 2-40 )

                                τ off = τ T 1 = (k isc ' )−1                          ( 2-41 )

                                  I on = k excφ fl E det                              ( 2-42 )

Here, φ fl and φisc are the quantum yields of fluorescence and the triplet state

transition, Edet is the overall detection efficiency, and τ T 1 is the spontaneous lifetime of
the lowest energy triplet state T1, which in turn is the inverse of the rate constant kisc’.
It has to be mentioned that this is only the case if there are no higher triplet states
involved as an additional photoinduced channel [English et al., 2000].

Autocorrelation Function

If count rates are lower or a higher time resolution is desired, the second order
autocorrelation function is the method of choice. Since no discrete and


distinguishable levels of intensity are required, a much lower time binning than in
duration histograms is possible, thus enabling the measurement of fluctuations
between molecular states occurring on a much shorter timescale.

Mathematically, the normalized second order autocorrelation function is defined as

                                              I (t ) I (t + t ' )
                            g ( 2) (t ' ) =                 2                  ( 2-43 )
                                                   I (t )

where I(t) is the observed fluorescence intensity. The intensity is shifted by a
parameter t’ and integrated and normalized over the measurement period.

The physical meaning of the autocorrelation as defined above is that it is proportional
to the probability to detect a photon at time t’ if there was a detection event at time
zero. This probability is composed of two basically different terms: The two photons
detected at time zero and at time t’ can originate from uncorrelated background or
from different fluorescing molecules and therefore do not have any physical
correlation (provided there is no interaction of the different fluorescing molecules).
These events will contribute to a constant offset of g(2)(t’) that is completely
independent on t’. Alternatively, the two photons originate from one and the same
molecule and are therefore physically correlated, leading to a time dependent
component of g(2)(t’). Thus, the temporal behaviour of the autocorrelation function is
solely determined by the correlated contributions of individual molecules. In many
experiments more than one process is involved, and sophisticated analysing
techniques of the obtained autocorrelation data allow to extract information on each
fluctuation originating from a different process [Enderlein et al., 2004]. Quantitative
information is available from amplitude data characterizing a fraction of molecules
fluctuating between two particular states.

Often applied to measurements in solution, this technique with its numerous
modifications is summarized under fluorescence correlation spectroscopy (FCS)
[Widengren and Rigler, 1998; Haustein and Schwille, 2003]. The autocorrelation
function allows determination of the mobility or diffusion characteristics of molecules
and hereby gives information about the particular mass of fluorophores or larger
biomolecules which are labelled with fluorophores. Furthermore, kinetic data on
additional transitions between other states, e.g. intersystem crossing, cis-trans


isomerizations or general fluorescence quenching pathways, can be derived using
this technique. A basic mathematical description is given by
                       g ( 2) (t ) = 1 +
                                      τ
                                          t      
                                                  1 + Ae − t / τ exp
                                                                        )        ( 2-44 )
                                          diff   

where N is the average number of molecules in the laser focus, τdiff is the diffusion
time and τexp is the time constant of a transition with an amplitude A that can be
described by exponential kinetics.

The advantage of FCS is the relative simplicity of the analysis. Its drawback is that it
works only within a very limited concentration range: If the concentration of
fluorescing molecules becomes too large (typically > 10-8 M), the contribution from
correlated photons from individual molecules, scaling with the number N of molecules
within the detection volume, becomes small compared with the contribution by
uncorrelated photons from different molecules, scaling with N2 [Enderlein et al.,
2004]. Another limiting factor in FCS measurements is that kinetic data cannot be
obtained beyond the diffusion time, i.e. the average time a molecule stays in the
detection volume, preventing measurements longer than a few milliseconds.
Furthermore, an overlay of diffusion events with other events of similar timescale
makes data analysis complicated.

The autocorrelation function, if applied to single-molecule fluorescence traces of
fluorophores immobilized on surface, is independent of the diffusion term and
removes the limited time window inherent to solution experiments. As a result, this
allows the analysis of fluorescence fluctuations on a much larger time scale, only
limited by the total measurement time, i.e. until a fluorophore is finally photobleached.
The remaining photophysical processes usually exhibit first order kinetics and can be
described by a sum of exponential functions of all i processes,

                                                     t         
                          g ( 2) (t ) = 1 + ∑ Ai exp
                                                                                ( 2-45 )
                                            i        i ,char   

where τ i,char denotes the characteristic time constant, and Ai the amplitude of the i-th

component. An autocorrelation function calculated for a solution measurement
compared to an autocorrelation calculated for a surface-immobilized molecule is
presented in figure 2-20. The figure demonstrates that transitions into longlived dark


states with a typical time constant larger than the diffusion time are not accessible by
solution measurements.

 Figure 2-20: (a) Autocorrelation of diffusing Atto488 molecules in a laser focus (excitation
 power of 1 mW) with a diffusion time of 0.56 ms and an exponential time of 9.41 µs. (b)
 autocorrelation of Atto647-DNA immobilized on protein surface, exhibiting two exponential
 time constants of 17.48 ms and 15.5 µs, respectively, obtained from the fluorescence trace
 shown in (c). The longer time component, attributed to the formation of a radical state, is
 not accessible in solution measurements.

As a result, g ( 2) (t ) can be used as a method to directly determine the lifetime of “off”
and “on” states and their corresponding amplitudes from fluorescence intensity
trajectories only. Under the assumptions that the autocorrelation is only examined at
times that are large compared to the fluorescence lifetime of the molecule and that
the integration times are smaller than the average on- and off-times, τ char of the i-th

component can be expressed as [Weston et al., 1998; Bernard et al., 1993]:

                              1            1          1
                                     =           +          = koff + kon              ( 2-46 )
                            τ char       τ off       τ on

                                                 τ off k on
                                         A≅           =                               ( 2-47 )
                                                 τ on k off

Similarly to duration histograms, on- and off-times for transitions into nonfluorescent
states can be determined in this way.

In this work, both approaches to derive kinetic data from the photon signature of
single emitters were used and complement each other. If molecules are immobilized
under aqueous conditions using a molecular anchor, the direct influence of chemical


changes on certain types of chromophores can be investigated quantitatively. It
hereby becomes feasible to observe the effect induced by changing pH or redox
properties, removing oxygen or adding triplet quenching substances.


Fluorescence resonance energy transfer, first discovered by FÖRSTER in the late
1940s, was used as a powerful tool to characterize interactions between
fluorophores, possibly attached to a larger molecule or more complex systems of
interest [Clegg, 1992]. In 1996, resonance energy transfer between a single donor
and acceptor chromophore was first realized by WEISS and co-workers, coined single
pair FRET (spFRET) or single molecule FRET (smFRET) [Ha et al., 1996]. As a
model system, they used two dyes, tetramethylrhodamine (TMR) and Texas Red,
attached to the 5’-ends of two complementary single stranded oligonucleotides with
an interchromophoric distance of 3.4 nm and 6.8 nm. After hybridisation of the DNA
sample, they obtained energy transfer efficiencies at the single molecule level of 85%
and 53%, respectively, which is in good agreement with ensemble data of the same
samples. An example of a
fluorescence intensity trace of a
single   FRET     system     with
Rodamine Green and TMR,
spaced by 3.4 nm, is shown in
figure 2-21.

In contrast to ensemble FRET
measurements, the efficiency of
energy transfer is determined in
a slightly different way, which is
caused by the digital event of
bleaching of either the donor or
                                      Figure 2-21: Fluorescence intensity versus time in
the acceptor molecule. In the         a smFRET system of Rhodamine Green and TMR,
case of acceptor bleaching, the       spaced by around 3.4 nm, recorded on two
efficiency E can be determined        spectrally separated detector channels.



                                       I D − I DA
                                 E=                                               ( 2-48 )
                                      I D − αI DA

Here, IDA and ID represent the donor fluorescence intensity before and after
photobleaching of the acceptor, respectively. The correction factor α (< 1) is
necessary because a photobleached acceptor may still absorb light. If α is zero, this
formula is equivalent to the conventional FRET equation (equation 2-19). Applying
equation 2-48 to figure 2-21 and the tolerable assumption that the donor fluorophore
does only emit on the blue channel, a FRET efficiency of 82% is determined.

If the donor molecule photobleaches first, the efficiency of energy transfer is
                                      I φ 
                               E = 1 + DA A 
                                    I φ                                         ( 2-49 )
                                       AD D 

where φA and φB represent the quantum yield of the single donor and acceptor.

In the following years, smFRET has become a common tool to investigate single
molecule dynamics on a large range of timescales and a huge variety of systems
[Deniz et al., 1999; Ha, 2001]. This can particularly be observed in biological
research, including stepwise rotation of an ATPase [Diez et al., 2004], protein
structure and dynamics [Wang and Geva, 2005], or fluctuations between open and
closed states of single nucleosomes [Tomschik et al., 2005]. Furthermore, smFRET
experiments are not limited to organic chromophores. Recently, the activation of the
Ras protein has been studied using YFP as a donor for smFRET [Murakoshi et al.,
2004]. Quantum dots begin to be used as well. [Hohng and Ha, 2005] Due to the
unique excitation spectrum of quantum dots, they may only be used as a donor
molecule, transferring energy to an organic chromophore or a fusion protein.

As a further development of FRET with two molecules involved, several approaches
to more-step FRET in ensemble measurements have been made in the past years
[Liu and Lu, 2002; Ohya et al., 2003]. At the single-molecule level, first results in
triple chromophore FRET have also been published recently [Hohng et al., 2004]. As
an extension to both spFRET and multistep FRET in ensemble studies, the aim of
this work was to realize a multistep FRET system and investigate this system at the
single-molecule level. In these experiments, it is advantageous to work with a
number of “orthogonal” properties of chromophores, since spectrally resolved
fluorescence information is not an unambiguous mean to differentiate energy transfer

steps. Pulsed excitation, different fluorescence spectra and fluorescence lifetime,
together with modulation of the polarization of excitation light are additional
parameters important to reveal single energy transfer steps. This is facilitated further
by stepwise building up of a multichromophoric system at the single molecule level,
which enables the observation of each energy transfer step separately.


3. Materials and Methods

3.1. Spectrally-Resolved Fluorescence Lifetime Microscopy

The main set-up built during the course of this work is a confocal line scanning
fluorescence microscope with laser light excitation. This set-up allows capture of
spectral and temporal information on single photons, detected from single
fluorophores. Depending on experimental requirements, the set-up has been
modified or extended, e.g. for time-correlated measurements at different laser
wavelengths or with polarized excitation. As it presents a central element in this work,
it will be discussed in detail in the following section.

A simplified scheme of the fundamental set-up for fluorescence imaging microscopy
with single-molecule detection efficiency is portrayed in figure 3-1.

   Figure 3-1: Principal set-up for spectrally-resolved fluorescence lifetime microscopy.
   BS: beamsplitter, EF: emission filter, APD: avalanche photodiode.

                                  Materials and Methods

Before going into details of a number of important elements of the set-up, a brief
summary about the working principle will be given in the following passage.

The core unit of the set-up consists of an inverted microscope (Axiovert 200M, Zeiss,
Germany) equipped with an x,y,z-piezo stage (PI-509, Physik Instrumente, Germany)
controlled via analog ouput computer cards (PCI-6713, National Instruments, USA).
The scanning range of the stage is 100 µm in x and y direction and 20 µm in z
direction, and the stage is directly connected to the microscope. A number of
different laser light sources, both continuous wave and pulsed, with wavelenghts
from 440 nm to 635 nm were used as excitation light source. If necessary, especially
if using broad band emitting diode lasers as light source, a narrow excitation filter in
front of the laser output was used. The laser beam was extended using a telescope
and coupled into the backport of the microscope, where it was directed into the
microscope objective (100x oil immersion PlanFluar, NA = 1.45, Zeiss, Germany).
The light beam was focused onto the sample and fluorescence light was collected by
the same objective, but separated in the detection path of the set-up using a dichroic
beamsplitter, which allows separating fluorescence light from shorter wavelength
excitation light. The parallel light beam was first focused onto a pinhole. To spectrally
resolve the incoming photons in the detection arm, three different dichroic
beamsplitters with appropriate cut-off wavelength were used. Finally, emitted photons
were focused onto the active area of four separate avalanche photodiodes (APD;
AQR-14 and AQR-16, PerkinElmer, USA). Photon arrival generates an electric signal
which is then processed by a time-correlated single-photon counting device (SPC-
630, Becker&Hickl, Berlin, Germany), which allows both continous mode operations
and time-correlated measurements. If working in the time-correlated mode, the
synchronization signal of a pulsed laser source supplies the reference for the SPC-
630 device and allows obtaining fluorescence lifetime information. Together with the
spectral information, this method is referred to as spectrally resolved fluorescence
lifetime imaging microscopy (SFLIM) [Herten et al., 2000].

In the set-up presented here, the confocal detection path has been realized applying
the so-called 2f principle. According to the lens equation for a lens with a focal length

                                   1 1 1
                                     = +                                         ( 3-1 )
                                   f  g b

                                 Materials and Methods

where g and b denote the object and the image length, respectively, both object size
and image size remain unchanged for g = b = 2f.

The whole set-up is synchronized by LabView (National Instruments, Austin, USA)
based software, allowing the piezo scan stage to be controlled on the one side and
attributing photon information obtained by the photon counting device to the
appropriate x,y-position on the sample.

The basic set-up presented in figure 3-1 has been extended by several components,
according to experimental requirements. Simultaneous excitation with two different
laser light sources were applied for dual colour measurements. Modulation of linearly
polarized laser light by an electrooptical modulator (EOM; Linos, Germany) allowed
for rotational dynamic studies of fluorophores conjugated to biomolecules. A charge
coupled device (CCD) camera (Cascade, Roper Scientific, USA) connected to the
sideport of the microscope equipped with a spectrograph was used for single
molecule spectra and alignment purposes.


Most experiments in this work were performed on
the Axiovert 200M inverted microscope (Zeiss,
Jena, Germany). This fully motorized microscope
(see figure 3-2) is equipped with a backport,
sideport, frontport and baseport, which facilitates
the construction of a multifunctional set-up. The
collimated laser beam enters the microscope via
the backport and is reflected by a dichroic mirror
into the objective. Additionally, a mercury lamp is
connected to the backport through a lense system,        Figure 3-2: Axiovert 200M, Zeiss,
and a software controlled flippable mirror allows
changing between excitation sources. Emission light collected by the objective
passes the dichroic mirror followed by a convex lens, and can now be detected on
either of the remaining ports. A simultaneous detection of 50% on both sideport and
baseport is also feasible.

                                     Materials and Methods

The detection pathway connected to the baseport of the microscope is realized by
applying a modified confocal scheme using two lenses, one inside the microscope (f
= 160 mm) and one in the detection arm (f = 300 mm). Spectral discrimination of the
emission light is achieved by using up to four avalanche photodiodes (APD) and
appropriate dichroic beamsplitters and filters.

Excitation Sources

Single-molecule fluorescence experiments were performed using laser light as
excitation source, exhibiting highly collimated and monochromatic light. Depending
on experimental needs, different types of lasers were used, that can be classified into
continuous wave and pulsed laser. An overview of both groups, together with
characteristic wavelengths and repetition rates, is given in table 3-1.

       Name                   Type           Wavelength         Power        Repetition Rate

                                                   (nm)          (mW)             (MHz)
95 SHG-6W              Argon-Ion Laser             514.5          200               cw
Lexel Laser Inc.,                                  496.5          30                 “
USA                                                488.0          100                “
                                                   476.5          10                 “
                                                   457.9          10                 “
LHYP-0201              Helium-Neon                  594           2.0               cw
Research Electro       Laser
Optics, USA
Polytec, Germany       Helium-Neon                 632.8          2.0               cw
Tsunami                Ti:sapphire Laser        740-935        650-1350         80.4 MHz
Spectra Physics,                                                              (80.77 MHz at
USA                                                                              976 nm)
Mira 900               Ti:sapphire Laser        680-980        200-1400         76.8 MHz
Coherent, USA                                 (980-1050)                       (at 976 nm)
LDH-P-C-440            Diode Laser                440              1           5 – 40 MHz
Picoquant,                                                   (at 40 MHz)
LDH-635                Diode Laser                 635            3.5          5 – 80 MHz
Picoquant,                                                   (at 80 MHz)
Table 3-1: Summary of laser systems used as excitation sources (see text for details).

                                  Materials and Methods

Due to low photostability of many fluorescent dyes absorbing in the blue and green
spectral region, most experiments had to be carried out using continous wave
excitation to avoid excitation of higher excited states and thus further destabilization
of the fluorophores. For these experiments, the continuous wave operating argon ion
laser served as excitation source, mainly operating at 488 and 514 nm. The laser
provided linear polarized light with narrow emission and <1.5 mm beam diameter. For
longer wavelengths that cannot be obtained from the argon ion laser, two continuous
wave helium neon lasers with 594 and 633 nm were used.

Time-resolved measurements reveal information about fluorescence kinetics and
depopulation processes of the excited state due to energy transfer or interactions
with the microscopic surroundings of fluorophores. Experiments were realized with a
number of different pulsed excitation lasers and are the subject of a following section.

Scanning Stage

As scanning stage, a three-dimensional closed-loop piezo scanner P-517.3CL
(Physikinstrumente, Göttingen, Germany) in combination with a three-channel
amplifier E503.00 (Physikinstrumente, Germany) and a capacitive sensor controlling
unit E509.C3A (Physikinstrumente, Germany) was used. The range of the scanning
device is 100 µm x 100 µm x 20 µm, controlled by a voltage ramp of 1 V / µm. To
address the scanning device, custom-made software (LabView Version 7.0, National
Instruments, USA) together with an analogue output PC-card (PCI-6713, National
Instruments, USA) was used, and allowed synchronized stage movement with data
acquisition of the fluorescence detectors.


Single-photon detection was realized using up to four avalanche photodiodes (APD;
sometimes referred to as single photon avalanche diode, SPAD; AQR-14 and AQR-
16, PerkinElmer, USA).

An APD is a solid-state, semiconductor based device that converts photons into
electrons in a different way than an ordinary photodiode. In ordinary photodiodes, the
photon/electron ratio is one to one. In APDs, however, photon-generated electrons
can excite additional electrons creating an avalanche effect and resulting in an

                                   Materials and Methods

internal gain that can be adjusted with the bias voltage, e.g. 1 to 1000 V. The
resulting signal is proportional to the light's intensity. The active area of APDs from
the AQR-1x-series has a diameter of about 180 µm. They are thermoelectrically
cooled and have a dark count rate of < 100 counts per seconds (cps) (AQR-14) and
<25 cps (AQR-16), respectively. The dead time between two subsequent photon
arrival events lies in the range of around 300 ns. The photon arrival time itself is
measured from the rise of the electric signal, generated by the APD, and can be
measured with a temporal resolution of 300 ps, an important factor contributing to the
instrument response function (IRF).

Besides avalanche photodiodes, other detector devices commonly in use for single-
molecule fluorescence microscopy are photomultiplier tubes (PMT) and charge
coupled device (CCD) cameras. The advantage of a PMT over an APD is a low dark
count rate (typically around 5 Hz) and a shorter dead time of around 10 ns. On the
other hand, a PMT suffers from poor quantum efficiency of around 20%. The
detection efficiency of an APD reaches its maximum at 700 nm with over 70%, and
the range of detection with more than 50% efficiency lies between 500 and 850 nm
(see figure 3-3). Best quantum efficiencies for photon detection are offered by CCD
devices, realizing more than 90 % in modern cameras. With frame rates in the kHz
regime, parallel image processing and straightforward alignment, CCD devices are
used more and more in many fluorescence microscopy applications.

Compared to a CCD device, an APD
works      on   a   considerably     faster
timescale, since a CCD can not avoid
readout and capacity charging times.
Lifetime    measurements     cannot     be
realized photon by photon, but require
summing up over longer timescales.
Thus, dynamic changes in fluorescence
lifetime at short timescales cannot be
observed.                                          Figure 3-3: Photon detection probability
                                                   for an APD of the AQR-1x series.

As a result, an APD represents the detector of choice for sensitive single molecule
experiments in a spectral range of 500 to 800 nm. A moderately high quantum yield

                                      Materials and Methods

together with a short temporal resolution and acceptable dark count rate is a well
suited compromise, and APDs have been used as detectors in all set-ups.

Filters and Dichroics

The information of a photon detection event can be refined by spectral discrimination,
which results in a valuable increase of information. Therefore, many experimental
set-ups operate with more than one detector and spectral resolution was achieved by
the use of dichroic beamsplitters and appropriate excitation and emission filters
(Omega Filters, Brattleboro, USA).

In the excitation path, a combination of a narrow band excitation filter and a dichroic
beamsplitter were used. An excitation filter can be set aside if the excitation laser is
characterized by narrow emission at exactly one wavelength, and no fundamental
wavelengths or other sources of light emission appear. This is particularly the case if
fundamental wavelengths of atom or gas lasers, e.g. continuous wave argon ion
lasers, are used. The following table 3-2 shows commonly used combinations of
excitation filters and beamsplitters

           Excitation Wavelength         Excitation Filter    Dichroic Beamsplitter

                   440 nm                   445 DF 20              465 DRLP

                   488 nm                   480 DF 40*             500 DRLP

                   635 nm                    633 DF 9              647 DRLP

             635 nm and 488 nm              633 DF 9**         Dual Band 635/488

Table 3-2: Combinations of dichroic beamsplitters and excitation filters used in fluorescence
microscope set-ups with laser excitation; (*) excitation filter is only necessary if frequency-
doubled titan sapphire laser are used, removing fundamental wavelength of infrared light; (**)
excitation filter was used for the red laser diode only.

Experiments with an excitation wavelength of 488 nm required a combination of
dichroic beamsplitter (500 DCLP) and long pass emission filter (500 ALP). The
transmission curves of both components are shown in figure 3-4 (left side). The
reason therefore lies in the property of band pass filters which were chosen for longer
wavelength detection channels and exhibited a small fraction of transmission at the

                                      Materials and Methods

 Figure 3-4: Transmission curves of excitation beamsplitter and long pass filter in
 experiments with 488 nm excitation (left) and dual wavelength excitation beamsplitter with
 excitation filter for a 635 nm laser diode (right) and high reflectivity at 488 nm.

excitation wavelength of 488 nm. Additional background signals on the detectors can
thus successfully be avoided by using a long pass filter before spectrally separating
emitted light onto the detectors. Experiments with dual-wavelength excitation were
carried out using a dual wavelength beamsplitter exhibiting high reflection at 488 nm
and 635 nm. A narrow band excitation filter (633 DF 9) has been used to spectrally
select laser diode emission (see figure 3-4, right side).

Spectrally resolved information on fluorescence photons has been realized by using
a set of dichroic beamsplitters and filters in the emission path of the set-up.
Configurations for two different excitation wavelengths are listed in table 3-3.

        Excitation Wavelength        Dichroic Beamsplitter             Emission Filter

               635 nm                                               665 DF 60, 675 DF 50
                                           680 DRLP
                                                                          700 DF 75

               488 nm                                                     525 DF 45
                                           560 DRLP
                                                                          580 DF 60
                                           600 DCLP
                                                                    645 DF 75, HQ 542 LP
                                           680 DRLP
                                                                          700 DF 75

Table 3-3: Beamsplitter and filters used in the emission pathway for 635 nm and 488 nm

                                   Materials and Methods

As an example configuration, transmission spectra for a set-up with 488 nm
excitation and four detector channels are shown in figure 3-5. In the lower part of
figure 3-5, the resulting spectral range for detection of each APD is depicted.

It is clearly visible that dichroic beamsplitters, e.g. the one shown at the left side in
figure 3-4, do unfortunately not exhibit constant transmission over a long range of

 Figure 3-5: Example of a configuration of beamsplitters and filters used with 488 nm
 excitation and four detector channels. Transmission spectra of filters (top) and dichroic
 beamsplitters (middle) are shown. The resulting spectral separation (bottom) has been
 calculated for each detector, ranging from 500 nm to 740 nm.

                                          Materials and Methods

wavelengths. Especially in experiments with a large spectral detection range, one
has to be aware of a decreasing transmission of the excitation beamsplitter occurring
at higher wavelengths, which have an important impact on spectral information. The
influence onto the detection pattern of the previously described four detector set-up
with 488 nm excitation is shown in figure 3-6. Additionally, the quantum yield for an
APD was considered, and total collection efficiencies are depicted by filled areas in
the graph. It is remarkable that the total detection efficiency is now reduced to an
average value of 30% of photons emitted.

Furthermore,                     some
beamsplitters exhibit strong
polarization           dependency,
depending on the fabrication
type. As a result, the rising
edge    in     the     transmission
spectrum can shift up to 20
nm,     according           to    the
polarization         properties     of
incoming light. It is therefore
advantageous to use dichroic
reflectors      (titled     DRLP),
characterized by a steep                 Figure 3-6: Comparison of detection patterns (dotted
transition     slope       and    low    curves) of a four detector set-up and corrected patterns
polarization dependence, if              (full curves) by considering an excitation beamsplitter
                                         (500 DRLP) and long pass emission filter (500 ALP). Filled
available      for     a    desired
                                         areas below take into account the detection efficiency of
wavelength. Dichroic filters
                                         APDs and yield total detection patterns.
(DCLP) on the other hand
provide wide regions of both transmission and reflection, but exhibit a high degree of
polarization along with a somewhat shallow transition slope.

                                    Materials and Methods


As a further extension to the basic set-up described above, time-correlated
measurements with four channel spectral separation were performed, using pulsed
excitation sources and a time-correlated single-photon counting (TCSPC) PC-plug-in
card (SPC-630, Becker&Hickl, Germany). Thus, excited state information in a
nanosecond timescale with minimum channel size of ~3 ps could be realised, and
allowed a further characterization of complex multichromophoric systems. The basic
changes in the set-up together with the working principle of the SPC-630 and the
characterization of the instrument is discussed in the following section.

Time-Correlated Single-Photon Counting (TCSPC) Using SPC-630

The computer based assignment of photons, both spectrally and temporal, was
carried out using a PC-plug-in card for time-correlated single-photon counting, SPC-
630 (Becker&Hickl, Berlin, Germany). Since this card represents the heart of data
acquisition, it shall be discussed in more detail. The basic structure is shown
schematically in figure 3-7.

 Figure 3-7: Schematic representation of the multichannel measurement principle, composed
 of a router unit and the PC-plug-in card SPC-630 for single-molecule photon counting.

                                  Materials and Methods

Briefly, photons detected by several single-photon counting devices, e.g. APDs or
PMTs, are spectrally encoded by attributing a routing bit to the temporal information,
done by the routing module. Together with a constant-fraction discriminator (CFD)
signal, responsible for an exact output pulse independent of the pulse height from the
detector, the routing signal is transferred to the SPC-630. Besides several other
functions, the CFD allows the reduction environmental noise or small background
pulses (particularly “after-pulses”, which are common if working with APDs) by
applying a threshold to incoming pulses from the detector.

The synchronization signal (SYNC) that is derived from the pulses of the light source
is essentially important as a trigger signal for the SPC-630 card, since the
microscopic times, i.e. fluorescent lifetime information, are calculated with respect to
the SYNC pulses. Comparably to the CFD discrimination capabilities, electronic after-
pulses from photodiodes or delay units can be corrected by adjusting SYNC

In the next step, the time-to-amplitude converter (TAC) is used to determine the
exact temporal position of a photon arrival signal with respect to the SYNC signal,
resulting in the microscopic time of a photon. Since the probability of detecting a
photon is considerably lower than detecting a SYNC signal, the TAC is usually
processed in the “reversed start-stop” mode, which avoids many cycles without
photon detection. In other words, after the arrival of a start signal generated by a
photon on a detector, the temporal position towards the next SYNC signal is
measured. To do so, the TAC generates a linear voltage ramp until the stop signal,
provided by the SYNC, is detected. Depending on adjustable TAC parameters, an
output voltage that depends linearly on the microscopic time is generated.

In a next step, the analogue digital converter (ADC) converts the analogue signal into
a memory address, MEM, by resolving it into 4096 time channels. In this step, a
maximum time resolution of ~3 ps can be achieved.

The SPC-630 module can be operated in continuous flow or in the first-in-first-out
(FIFO) mode. Processed in the continuous-flow mode, photon histograms are
recorded on board using two available memory banks and hereby preventing read-
out gaps during the measurement. This mode is strictly hardware based and thus
provides an extremely accurate recording sequence. Alternatively, a FIFO structure
of the memory of the card can be used. In this mode, the measurement does not

                                  Materials and Methods

      Figure 3-8: SPC-630 card working in the FIFO-mode.

deliver a histogram but a continuous stream of information about the individual
photons. The principle is shown in figure 3-8.

Additionally to the signal created by the ADC, a macro timer delivers the macroscopic
time since the beginning of the measurement. The resolution of the macro timer of
the SPC-630 is 50 ns. By this, each photon is encoded by the routing information, a
microscopic time and a macroscopic time, together with some control flags to
process overflow, invalid photons or other events. These data are bundled into one
FIFO unit, which, depending on the time resolution chosen, has a size of either 32 or
48 bits. Software can continuously read out the FIFO buffer after a certain number of
photon arrival events, and time resolved information on fluorescence data is now
available, photon by photon.

The fundamental advantage of this method is that each photon is fully described by
its FIFO package. Hence, a powerful and multidimensional data analysis becomes
possible, from fluorescence intensity time traces including spectral information,
fluorescence lifetime and coincidence of photons. Additionally, photons can be
histogrammed or correlated, which allows the identification of kinetic information on
both a microscopic and a macroscopic timescale.

If the set-up is operated in a multichannel mode and separate microscopic times for
each channel are of interest, the response from each detector channel is not
necessarily in the same time window and has to be synchronized. This is caused by

                                 Materials and Methods

unequal detection pathway lengths of different channels by variable cable lengths
that connect the detectors’ signal with the routing unit. To circumvent the loss of
photons and assure the accuracy in such measurements, one channel is chosen as a
reference signal. Further detector channels are overlaid in time using three delay
units with 1 ps time resolution (DG535, Stanford Research Instruments, USA

Pulsed Laser Excitation

If particularly fluorecence kinetics and not photostability of fluorophores were of
interest, different pulsed laser systems were used as excitation source. Due to
increased photostability of fluorescent dyes in the far red, these dyes are available
for excitation with a pulsed source, namely a diode laser with a centre wavelength of
635 nm (LDH-635, Picoquant, Berlin, Germany) and a pulse width between 100 and
300 ps, depending on the laser output power. A picosecond pulsed diode laser driver
(PDL-800-B, Picoquant, Berlin, Germany) was used for pulsing the laser diode with a
base frequency of 80 MHz, which could be adjusted dividing by constant factors of 2,
4, 8 or 16. Since the emission profile of diode lasers is usually broader and
sometimes slightly shifted in maximum intensity, a narrow band excitation filter was
used (e.g. 639DF9, Omega Optics, USA) to further narrow down the excitation

Pulsed excitation at shorter wavelengths was realized with three different laser
systems, mainly distinguished by the operating wavelength, pulse width and
repetition rate. A peltier cooled laser diode with 440 nm centre wavelength (LHD-P-
C-440, Picoquant, Berlin, Germany) and a pulse width ranging from 90 to 300 ps was
used in the lower spectral range, but is only suited for excitation of poorly stable
coumarine dyes. To access the excitation of rhodamine dyes which are generally
preferred to coumarines in single molecule experiments, and to allow comparable
conditions to continuous wave excitation using the 488 nm line of an argon ion laser,
two different frequency doubled titan sapphire lasers were used. Pulsed light with a
centre wavelength between 480 nm and 490 nm was provided, exhibiting a pulse
width of 150 fs (MIRA 900, Coherent, USA) and 130 fs (Tsunami, Spectra Physics,
USA), respectively. Due to extremly short pulses and hence high pulse energy
densitites of femtosecond titan sapphire systems, the frequency doubled Tsunami
laser was coupled into a single mode optical fiber (3.2 µm core; S405-Custom,

                                     Materials and Methods

Nufern, USA) of 50 m length. The broadening of the pulse by dispersion in the fiber
could be estimated by

                                              4 ln 2          
                         ∆τ out = ∆τ in   1+             β 2 L                       ( 3-2 )
                                              (∆τ in )
                                                              

where ∆τ in represents the input pulse width, ∆τ out the output pulse width, L is the

length of the fibre, and β 2 is the group velocity dispersion (GVD) of the fibre material,
depending on the wavelength. With β 2 = 70 ps2 km-1 and ∆τ in = 130 fs, a pulse width

of around ∆τ out = 100 ps is obtained. To verify this result experimentally, a fast

photodetector (Antel Optronics, USA) with a spectral range of 300 to 1100 nm and a
FWHM pulse width of < 65 ps, together with a fast sampling oscilloscope (Tectronix
7904, Oregon, USA) with a time constant of 25 ps have been used. The oscilloscope
response is shown in figure 3-9 (left side) and shows a broadening of around 160 ps
for the FWHM of the observed pulse, which exhibits a tailing of up to 500 ps. This is
comparably to the theoretically obtained result, if one takes into account that the
measured pulse is convoluted with the detector and oscilloscope characteristics. The
spectral distribution of the frequency doubled laser wavelength was determined to be
5 nm (figure 3-9, right side).

 Figure 3-9: left: response function of a Tsunami laser pulse after 50 m single mode fibre,
 measured with a fast oscilloscope, pulse width of 160 ps (one grid represents 100 ps); right:
 spectrum of the same laser centred at 487 nm, and a FWHM of 5 nm.

                                           Materials and Methods

Instrument Response Function

The instrument response function (IRF) of a set-up which is operated with pulsed
excitation represents the bottleneck for the accessible time resolution. Typically in the
same order of magnitude as fluorescence kinetics from organic chromophores, the
accuracy of data analysis depends strongly on the width of the IRF.

Theoretically, an IRF is described by the convolution of all steps influencing the pulse
width, both from the excitation path and the emission path. Here to mention are the
pulse width of the laser source as well as the time resolution of the detector elements
and the TCSPC plug-in cards. Some of the components of an IRF can be measured
independently, as shown for the pulse width of the Tsunami laser previously. Others
depend strongly on experimental conditions, as the time resolution of an APD
correlates with the fraction of the active area which is used to detect photons.
Assuming a laser pulse width of 160 ps, a temporal response of 200 ps for the PC
plug-in card and 300 ps as best response of an APD, the theoretically best value for
an IRF can be estimated to be around 350 ps, using the following estimation:

                                                 ∑ (FWHW )
                                FWHWIRF =                     i                                 ( 3-3 )

where FWHWi represent the full width half mean of the IRF of the single components,
yielding the total IRF, denoted

The total IRF for each detector
channel of the Tsunami set-up
described in this section has
been previously determined to
have a FWHM of between 1.0
and 1.2 ns, shown in figure 3-10.
Main reasons for this significant
broadening of the IRFs are
derived from bandwidth of the
                                               Figure      3-10:       Instrument   response   functions
SYNC       signal   of        the     laser,
                                               measured for each detector of the Tsunami set-up,
provided      by         an         internal   exhibiting a FWHM between 1.0 ns and 1.2 ns.

                                  Materials and Methods

photodiode, as well as each individual APD, whose resolution function depends on
the effectively used fraction of the active area. This is particularly important if a 2f
detection path is used, which is often poorly sensitive in the z-direction and produces
considerably larger images on the detector.

Instrument response functions of around 1 ns are a limit to time resolution, and
fluorescence decays of around 0.5 ns are the shortest which can be observed. In
experiments with efficient depopulation of a fluorophore’s S1 state by e.g. energy
transfer or quenching, fluorescence decays of this magnitude are generally observed.
Therefore, deconvolution of fluorescence decays of each channel with the
corresponding IRF have been applied, which allow a theoretical time resolution of
around 1/20 of the FWHM of the IRF, i.e. 50 to 60 ps.

To substantially improve the time resolution, a modified detection scheme with
parallel optical paths to all detectors, finally focussing onto a small fraction of the
active detector area with a lens of short focal length, can be applied. Experiments on
a set-up designed with this detection principle have been carried out using the
MicroTime 100 system (PicoQuant, Berlin, Germany). A brief description is given in
the following section.

                                   Materials and Methods


Additional to the set-up mentioned above, time correlated measurements were
performed on the MicroTime100 set-up (Picoquant, Berlin, Germany) containing four
APDs and a 470 nm laser diode as excitation source, pulsed with 40 MHz. The
schematic principle is shown in figure 3-11, and a more detailed description is the
subject of this paragraph.

   Figure 3-11: Schematic view of the Microtime100 set-up (Picoquant, Berlin, Germany). A
   modified set-up with four detector channels was used in this work.

The MicroTime100 set-up uses an IX71 (Olympus, USA) inverted microscope and a
similar x,y piezo stage to the one previously described. The excitation laser is
coupled into a fibre, enters the optical part of the set-up and is directed into the
sideport of the microscope via a dichroic mirror. Fluorescence light passes through
the same sideport, transmits the dichroic mirror and is spectrally resolved onto four
detector channels using the same dichroic beamsplitters and filters in the detection
path as described for the basic set-up.

The most striking difference is located in the detection pathway of the optical part of
the set-up. In earlier described set-ups, light leaves the microscope already focused
by a lens positioned directly after the excitation dichroic beamsplitter, is then
projected onto a pinhole and finally on the active area of a detector. In this particular
                                            Materials and Methods

set-up, parallel light leaves the microscope, and the confocal pinhole is placed
between two lenses of a telescope. Subsequently, the light beam is parallel again
until reaching the desired detector, which allows the length of the detection path to be
independent and gives more freedom in designing a detection path. Furthermore, all
optical elements like filters or beamsplitters do not affect the quality of the image and
can be exchanged easily. A lens with a short focal length, which can be adjusted in
all three dimensions, is finally used to focus light onto the detector. Easy adjustment
of the detectors is achieved, and using a small area in the centre of the active area
has been experimentally proven to have a positive effect onto the FWHM of the
instrument response function.

Similar to the SPC-630 card used in previously described set-ups, the MicroTime 100
is processed by a TimeHarp 200 PC-plug-in card. The operation mode is similar,
although titled time-tagged time-resolved (TTTR), with a time resolution of <40 ps.

Again of particular interest to
estimate the quality of the
nanosecond time resolution, the
total IRF of this set-up was
determined for all four detector
channels      independently.         The
graphical        representation       is
shown    in      figure    3-12.     The
FWHM        of    the     IRF   of    all
channels was determined to be
around 0.7 ns each, which is                  Figure 3-12: Total IRF measured for all four detector
around 0.3 ns shorter than for                channels of the MicroTime 100 set-up independently.
the Tsunami set-up described

                                    Materials and Methods


In experiments aiming to probe a single chromophore with two different wavelengths,
two laser beams were overlaid using a dichroic beamsplitter before entering the
backport of the microscope. A schematic representation of two-colour excitation is
presented in figure 3-13. First, both lasers, exhibiting different beam diameters, were
extended using a telescope, and two beams with identical diameter were obtained.
By using a dichroic beamsplitter (658 DRLP, Omega Optics, USA) which shows high
reflectivity for 635 nm and sufficient transmission around 488 nm, the laser beams
were overlaid.

To assure the best possible overlay of the excitation volumes of both foci, one
telescope was equipped with an adjustment along the optical axis. Using a CCD
device and the electronic movement of the objective provided by the microscope, the
focal volumes could hereby be overlaid with an accuracy of more than 0.5 µm along
the optical axis.

     Figure 3-13: Schematic principle of two-colour excitation. A dichroic beamsplitter
     (658 DRLP, Omega Filters, USA) reflecting light at 635 nm and transmitting light in
     the region of 500 nm was used to overlay different laser beams.

                                    Materials and Methods

3.2. Ensemble Spectroscopic Instrumentation

Ensemble spectroscopic measurements can be divided into steady-state and time-
resolved measurements. Although suffering from averaging of subpopulations and
poorer sensitivity, ensemble methods are indispensable for characterization
purposes of fluorophores and modified biomolecules. These standard techniques
served to characterise complex multichromophoric constructs which further are
investigated in single-molecule experiments. A short overview is given in the
following passages.


Absorption Spectroscopy

Absorption measurements were performed on two UV/VIS spectrometers, a Lambda
25 spectrometer from Perkin Elmer (USA) and a Cary 500 spectrometer from Varian
(Darmstadt, Germany). Samples were measured in silica precision cuvettes (suprasil)
from Hellma (Mülheim, Germany). Depending on experimental conditions, both semi-
micro cuvettes with a path-length of 10 mm and micro cuvettes with a path-length of
3 mm were used.

To avoid intermolecular interactions in absorption measurements and fluorescence
emission re-absorption effects, the concentration of samples did not exceed 10-6 M.
In this case, the LAMBERT-BEER law is valid,

                                A = log            = ε (λ )cd                  ( 3-4 )
                                          I (λ )

where A is the absorption defined as the decadic logarithm of the fraction of the
reference intensity, I0, and the measured intensity, I(λ). If the absorption is below a
value of around 1, linearity to the concentration, c, of the sample is observed,
normalized by the path-length of the sample, d, and the wavelength dependant
extinction coefficient, ε(λ).

                                   Materials and Methods

Fluorescence Spectroscopy

Fluorescence spectra were obtained using a Cary Eclipse spectrometer from Varian
(Darmstadt, Germany). Besides fluorescence intensity measurements, the instrument
offers the possibility to obtain dynamic fluorescence measurements at different
emission wavelengths simultaneously. Furthermore, polarized excitation and
detection is possible, enabling anisotropy measurements. A temperature control unit
based on a thermoelectric peltier device allows measurements of thermal profiles
between 4 and 90 °C.

Briefly, the spectrometer uses a xenon flash lamp technology providing an excitation
range between 290 to 1100 nm. A photomultiplier tube (PMT) is used as detector,
and sample concentrations down to 10-9 M can be measured.


Ensemble fluorescence lifetime measurements were carried out using the technique
of time-correlated single photon counting (TCSPC) with an IBH spectrometer (model
5000MC; Glasgow, Scotland). As excitation sources, various pulsed light emitting
diodes (LED) with a central wavelength of 450 nm, 495 nm and 590 nm and a pulse
width of around 500 ps were used. For excitation at 635 nm, a laser diode with a
pulse width of around 50 ps was used. The repetition rate is set to 1 MHz. Photon
detection is realized by a photomultiplier tube (PMT), sensitive for single-photon
detection. For the purpose of time-resolved anisotropy measurements, both
excitation and emission polarization could be modified.

Lifetime histograms were taken by collecting a fixed number of photons in the
maximum channel, e.g. 2000 to 5000 photons, depending on the sample
concentration. Up to 4096 time channels with a minimum bin width of 12 ps were
used. A fluorescence decay of a DNA bound fluorophore upon excitation with 450 nm
is shown in figure 3-14, together with the pulse profile of the excitation source, i.e. the
lamp profile or “prompt” signal.

                                      Materials and Methods

Figure 3-14: Fluorescence decay of Alexa430 conjugated to the 5’-end of a 60 base pair double
stranded DNA, obtained by excitation with 450 nm LED (pulse of LED in gray). Fluorescence
lifetimes were calculated after deconvolution and applying a biexponential fit.

As the excitation pulse is not infinitesimal short, a convolution method was used to
obtain exact fluorescence lifetimes of multi-exponential decays with up to four
components. The mathematical routine was part of the spectrometer’s software
package “Data Station 2000”. Briefly, a set of start parameters is defined, followed by
the calculation of a fluorescence decay by a deconvolution and fitting procedure.
With the software provided, the decay in figure 3-14 was fitted to two components of
3.58 and 6.00 ns. An estimation of the quality of a fit is given by a normalized χ2
value calculated by the software. As a rule of thumb, χ2 values better than 1.20 give a
confidence level of 90% for the calculated fit, and best fits are obtained with χ2 values
around 1. The bi-exponential fit in figure 3-14 has a χ2 value of 1.08.

As an extension to fluorescence lifetime measurements at only one detection
wavelength, the spectral dependence of a fluorescence decay is accessible by time-
resolved emission spectroscopy (TRES). This is of particular interest if a
depopulation process of S1 of a fluorophore is influenced by the presence of a
second fluorophore within short distance, which is the case in fluorescence
resonance energy transfer experiments. If multichromophoric systems are subject to

                                    Materials and Methods

investigation, time resolved fluorescence measurements can help to unravel the
efficiency of single transfer steps, additionally to spectral intensity changes.

3.3. Biological and Chemical Methods


A broad spectrum of reactive organic fluorophores are commercially available. Two
main classes of chemical modification are used frequently, which are amine-reactive
dyes containing a succinimidyl ester (NHS ester) group, and thiol-reactive dyes
containing a maleimide ester group. The underlying chemical reactions are depicted
in figure 3-15.

In short, a nucleophilic substitution at the carboxyclic ester group (case A in figure 3-
15) initiates the conjugation of an amine-reactive fluorophore (R1) with R2, the
molecule of interest, e.g. an amino modified DNA. The reaction is driven by the
formation of an energetically stable N-succinimide. A thiol-reactive probe undergoes
a nucleophilic addition of a thiol-containing molecule (e.g. cysteine residues from
proteins) at the double bond of the maleimide (case B in figure 3-15). The

 Figure 3-15: Chemical reaction scheme for labelling biomolecules. (A) succinimidyl ester of
 fluorophores are conjugated to amino functionalised molecules, (B) maleimide ester are
 conjugated to thiol groups.

                                     Materials and Methods

nucleophilic substitution of NHS esters is catalysed by slightly alkaline conditions,
whereas maleimide esters react in neutral environment.

If the molecule of interest has to be labelled with two different fluorescent probes at
well-defined sites, both strategies are applied and the selectivity of each reaction
type results in a doubly labelled molecule.

Amino modified oligonucleotides (purchased from IBA GmbH, Göttingen, Germany)
were the molecular basis for many samples generated in this work. A large number
of fluorescent probes from different companies, summarised in table 3-4, were
conjugated to oligonucleotides of different length.

  Company                                      Fluorophores

  AttoTec GmbH, Siegen, Germany                Atto590, Atto620, Atto647 and others

  Amersham Biosciences, NJ, USA                Cy3, Cy3.5, Cy5, Cy5.5

  Molecular Probes, Eugene, USA                Alexa488, RhodamineGreen and others

  Sigma Aldrich, St. Louis, USA                HIDC

  Denovo Biolabels, Münster, Germany           Oyster 656 and others

Table 3-4: List of manufacturers of different types of chromophores, used in this work.

Activated fluorophores are provided as dry powder and are only poorly stable in
water. Therefore, the dyes are usually diluted in DMF, and aliquots are stored at –
20 °C.

Dye-Labelling and Purification of DNA Oligonucleotides

Using post-labelling techniques together with hybridisation, synthetic oligonucleotides
offer a strong basis to construct multichromophoric compounds with defined
structure. Many modifications of oligonucleotides, including amino modification or
biotin tags as molecular anchor, are available and make DNA to a well suited
candidate as a molecular building block system. By using several independently
labelled   single   stranded    oligonucleotides,     various    combinations     of      a   final
arrangement are possible.

                                  Materials and Methods

In the frame of this work, oligonucleotides of different length between 20 and 60
bases, modified with amino groups and biotin as anchor molecule were used. Post
labelling and a multi-step purification by gel-filtration and high performance liquid
chromatography (HPLC; 1100 series, Agilent, USA) ensured purity of samples.
Labelling of single strand oligonucleotides with NHS-esters of various fluorophores
was realized using the following protocol:

                    10 nmol      Oligonucleotide (solution, ca. 10-4 M)

                    40 µl        Phosphate buffered saline (PBS; Sigma-Aldrich))

                    10 µl        100 mM carbonate/bicarbonate buffer, pH=9

                    30 nmol      Dye, diluted in DMF (Fluka)

After shaking for 1 – 2 h of, the sample was loaded onto a gel filtration column (NAP-
10, Amersham Biosciences, USA), previously conditioned with PBS. This purification
step allows a first separation step of DNA and excess dye by size exclusion.
Molecules with a molecular mass below ~1500 g/mol are retarded, while larger
molecules pass through the column without hindrance. Further purification was
realized by HPLC, using a reversed phase C-18 column and a 20-minute gradient
between two solvents A and B with following compositions:

        Solvent A (aqueous)      0.1 M Triethylammonium acetate (TEAA) in water

        Solvent B (organic)      0.1 M TEAA in 75% acetonitrile, 25% water

To remove solvents used in the HPLC purification step, the obtained solution of dye-
labelled DNA was in vacuum. If the fluorescent dyes used are not stable at alkaline
conditions, another gel filtration step had to be applied to replace the solvent mixture
by PBS or water. Otherwise, the pH during this step reaches strongly alkaline values
due to up-concentration of TEAA.

                                  Materials and Methods

Hybridisation of Single Strand Oligonucleotides

Concentrations of dye-modified oligonucleotides were derived from steady state
absorption spectroscopy as described previously. Since the extinction coefficient of a
fluorophore might be slightly changed if attached to an oligonucleotide, the sample
concentration was calculated using the absorption of both the fluorophore at its
maximum and the DNA itself, absorbing at 260 nm.

Purified and dye-labelled single-strand oligonucleotide fragments could now be
mixed, and thermal hybridisation was realized using a programmable thermo
controller (PTC 100, MJ Research, USA). To surely exceed the melting temperature
of DNA (which can be estimated to be around 70 °C), the samples were heated
shortly to 95 °C, cooled down to 65 °C in a minute, and slowly cooled down to room
temperature for about one hour.


To observe single fluorophores for a longer period of time, i.e. until photobleaching,
they have to be immobilized on a surface. This can generally be realized by different
techniques, either in dry or liquid environment. Throughout this work, two different
approaches to prepare samples with immobilized fluorophores were used: molecules
adsorbed on dry glass surfaces or dispersed into polymer films on the one hand, and
molecules attached to a protein surface using an anchor strategy on the other hand.

Dry Glass Surface

Cover slides (Roth, Karlsruhe, Germany) with a thickness of ~170 µm were used to
adsorb dye molecules. Before use, cover slides were treated with a 0.5% solution of
hydrofluoric acid in water. Afterwards, a solution of 0.1% of 3-aminopropyl-
triethoxysilane (APS; Sigma-Aldrich) in methanol was added and washed off with
water after five minutes incubation. The surface treatment results in a slight
adsorption of negatively charged DNA molecules onto a positively charged surface.
Samples for single molecule experiments were prepared by incubating the cover

                                 Materials and Methods

slides with a 10-10 M solution of DNA-dye-conjugates for several minutes. After a
washing step with water, samples were dried under nitrogen flow. This treatment
avoids contamination with impurities and yields the desired areal density for single
molecule experiments on a surface, which is around 1 molecule per square

Polyvinyl-alcohol (PVA, Sigma-Aldrich) as a polymeric host matrix was used to
prepare samples of fluorophores in an oxygen-free environment. The polymer was
dissolved in water at a concentration of 1% (w/v) by stirring and heating in a water
bath. Fluorophores were embedded in the polymer solution and coated onto a cover
slide surface using a spin coater, rotating the sample with 2500 rounds per minute.
Hereby, the solvent was removed and a homogeneous polymer matrix was obtained.

Immobilizing Molecules Under Aqueous Conditions

To allow the observation of biomolecules in a homogeneous environment,
measurements under aqueous conditions are the method of choice. A further
advantage lies in the possibility to change the chemical environment rapidly, and
hereby investigate the direct influence on a the photophysical behaviour of
fluorophores. In this work, a protein based anchoring technique which exploits the
binding of biotin molecules with the protein streptavidin was used [Anzai et al., 1995;
Dohnta et al., 1997].

To prepare the protein surface, cover slides with eight chambers (LabTek, Nunc,

                                                         Figure    3-16.     Immobilization   of
                                                         fluorophores using the binding
                                                         protein streptavidin. A surface
                                                         coated with BSA and BSA-biotin is
                                                         treated          with      streptavidin.
                                                         Fluorophores            attached     to
                                                         oligonucleotides containing biotin
                                                         as anchor bind to streptavidin.
                                                         The      lower     part    shows     an
                                                         increasing       number    with    time,
                                                         immobilized at the surface and
                                                         tracked online.

                                 Materials and Methods

Karlsruhe, Germany) glued to a glass slide were used. The surface was shortly
treated with 0.5% hydrofluoric acid solution (less than one minute), washed with
water, and incubated for several hours with a solution of 5 mg/ml of bovine albumine
serum (BSA, Sigma Aldrich, USA) and 1 mg/ml of biotinylated BSA (ca. 13 mole
biotin / mole BSA; Sigma Aldrich) in PBS. Prior to measurements, the BSA solution
was washed off the chambers several times with PBS. Subsequently, the surface
was incubated with a solution of 0.1 mg/ml streptavidin (recombinant, assuring
efficient binding of at least two biotin molecules per streptavidin molecule; Roche,
USA) for a couple of minutes. After application of washing steps, surfaces were
ready for use.

On the scanning stage of the microscope, the surface was loaded with a sample,
which was usually a fluorophore attached to an oligonucleotide having a biotin linker
at one of its ends. A few microlitres with a concentration around 10-9 M were added,
and the “loading” of the surface was tracked online. After a few minutes, the desired
density of molecules at the surface was reached. To remove remaining sample
molecules in the solution, the chamber was washed a several times with PBS. Figure
3-16 portrays this immobilization method and shows the increase of fluorophores
attached to the surface with time.

Using this technique, fluorophores attached to oligonucleotides rotate freely in
solution, which can be proved by rotating the polarization of the excitation source.
Furthermore, single stranded oligonucleotides anchored to the surface are accessible
for hybridisation, if a complementary DNA strand with slightly higher concentration is
added into the chamber. Altogether, this technique allows for higher flexibility in
single-molecule experiments, and opens many possible ways of experimental


The photophysical properties of fluorophores strongly depend on the environmental
conditions. The heterogeneity of photophysical parameters can be reduced if a
homogeneous microenvironment is realized. In other words, the distributions of
parameters as fluorescence lifetime and emission wavelength are broader if
fluorophores are adsorbed to a glass surface, compared to molecules immobilized
under liquid conditions using the anchor strategy. This can be explained by a

                                  Materials and Methods

randomly oriented adsorption of fluorophores and the lack of mobility on a glass
surface, resulting in different conformations and “nano-environments” for each
molecule, which alters fluorescence properties. Anchoring fluorophores in liquid
environment circumvents these effects and additionally offers the possibility to
change the chemical microenvironment easily. Especially the oxygen concentration
and redox properties of the surrounding solvent have been found to be important for
photophysical properties of organic fluorophores. The strategy to influence these
parameters is outlined in the following section.

Oxygen Removal

Efficient removal of oxygen was realized using an enzymatic approach [Funatsu et
al., 1995]. A solution of oxygen consuming enzymes, i.e. glucose oxidase and
catalase, was added to the solvent covering the fluorophores attached to the protein
surface. The following protocol was used to prepare 1 ml of enzyme stock solution

               1 mg Glucose oxidase (Sigma-Aldrich)

               0.5 ml Tris buffer (Roth, Germany), 25 mM KCl

               2 µl Catalase 30 mg/ml (Sigma-Aldrich)

               4 µl 1 M Dithiothreitol (DTT, mol. biology grade; Sigma-Aldrich) in

               0.5 ml Glycerol (GC-grade, Sigma-Aldrich)

Fractions of 50 µl were stored at –20°C. Prior to use, one fraction was diluted in 1 ml
of a 0.1 g/ml glucose (Sigma-Aldrich) solution in PBS. About 500 µl of this mixture
were transferred into a LabTek chamber. To reduce oxygen exchange with the
surrounding air, a “Press-To-Seal” silicone plastic (Molecular Probes, USA) was
used, covering the chamber.

In a first step, glucose oxidase requires glucose as substrate and consumes oxygen
to oxidize the substrate forming hydrogen peroxide. In a second step, catalase
reduces hydrogen peroxide to water.

                                      Materials and Methods

Enzymatic removal of oxygen is an efficient technique and decreases the
concentration of oxygen with a comparable efficiency to other methods, e.g.
replacing of oxygen by nitrogen flow or embedding molecules into PVA polymer
matrices. As a nice advantage, these proteins only slightly increase the background
intensity and make their application in single-molecule experiments possible.

Redox properties

Populations of the radical state of fluorophores can be manipulated by changing the
redox properties of the solvent. Fluorophores are either quenched by an electron
transfer reaction, or recover fluorescence if they are currently in a nonfluorescent
radical state. Furthermore, molecular switches exhibiting fluorescence states
depending on the redox environment have been demonstrated [Yan et al., 2005]. An
overview of chemicals which generate reducing or oxidising conditions is given in
table 3-5.

        Reducing reagent                                 Supplier

        β-mercaptoethanol (BME)                          Fluka

        2-Mercaptoethylamine (MEA)                       Fluka

        L-cysteine                                       Fluka

        Tetrachlorobenzoquinone (TCBQ)                   Sigma-Aldrich

        L-tryptophane                                    Fluka

        Guanosine monophosphate (dGMP)                   Sigma-Aldrich

        Oxidising reagent                                Supplier

        NaCNBH3                                          Sigma-Aldrich

        Ascorbic acid                                    VWR Prolabo

        Hydrogen peroxide (H2O2, 30%)                    Fluka

Table 3-5: Summary of reducing and oxidising reagents used to manipulate single
fluorophores in liquid environment.

                                  Materials and Methods

Both tryptophane and dGMP interact with some fluorophores by transferring an
electron via the mechanism of PET [Sauer 2003; Neuweiler et al., 2002;]. This
process of oxidation leads to static or dynamic quenching of fluorescence and can be
exploited for mechanistic studies of small distance changes or binding studies. A
prominent fluorophore which exhibits strong interaction with tryptophane by ground-
state complex formation is MR121 [Doose et al., 2005].

One widely used group of reducing chemicals contains a thiol group. These reagents
have an appropriate energetic potential which leads to electron transfer to some
classes of fluorophores, e.g. rhodamines. As a further property, these reagents
exhibit significant quenching of energetically appropriate triplet states of fluorophores
[Widengren and Schwille, 2000]. In addition, a reductive environment is a necessary
prerequisite for carbocyanine switching studies presented later in this work
[Heilemann et al., 2005; Bates et al., 2005].

4. Results and Discussion

4.1. Photophysical Properties of Fluorescent Dyes

To realize functional multichromophoric systems and to allow stable observation of
single systems under well-defined conditions over an extended period of time, a
number of considerations has to be made. First, different classes of fluorescent dyes
have to be explored under various conditions. Single-molecule experiments were
carried out both on dry glass substrates and under aqueous conditions. A
comparison of both environments should elucidate the influence of inhomogeneous
broadening on fluorescence properties of chromophores. Dyes of different chemical
classes were subject of a systematic investigation in this first part of this work. By
reducing photodestruction caused by laser induced irreversible reactions of the
chromophore and preventing all types of chemical destruction, a maximum of
fluorescent photons can be obtained.

Fluorophores suitable for single molecule experiments have to fulfil a number of
requirements. They must show a high quantum yield, high photostability to allow the
detection of as many photons as possible before irreversible photobleaching, and low
rates of parallel depopulation pathways of the first single state S1, i.e. low triplet
quantum yield and poor tendency to form radical states.

Furthermore, possible interactions with other molecules which influence the
photophysical behaviour become more and more interesting. The study and
prevention of possible photophysical pathways can result in a better stability. As a
prominent example, it has been shown that carbocyanine dyes which exhibit
energetically favourable interactions with molecular oxygen and hereby depopulating
the triplet state, can be stabilized by removing oxygen and adding millimolar
concentrations of a triplet quencher [Funatsu et al., 1995; Ha et al., 2002]. This
method is applicable to a large number of carbocyanine dyes, spanning an important
part of the visible spectrum. Today, this class of dyes is widely used in single pair
FRET experiments, and oxygen-free environments allow studies of hundreds of
seconds without photobleaching events [Hohng et al., 2004; Rasnik et al., 2005].

                                  Results and Discussion


A number of different classes of synthetic organic fluorophores are commercially
available. They can be further classified by their solubility in solvents, absorption and
emission wavelength, fluorescence lifetime and fluorescence quantum yield. The
choice of a particular type of chromophore first depends on the desired wavelength
for an experimental application. Rhodamine dyes span the visible spectrum from an
absorption wavelength of ~500 nm to ~800 nm. Unfortunately, rhodamine derivatives
with a functional group for conjugation chemistry are only available up to 650 nm. As
an alternative for the red part of the visible spectrum, a class of fluorophores which is
derived from the chemical structure of rhodamines and exhibits longer absorption
and emission wavelengths is available with carborhodamines. Structurally, they are
generated by replacing the central oxygen-atom by a methylen group. It will be
demonstrated in the following section that they exhibit similar properties, which
justifies to adjoin carborhodamines to the class of classical rhodamine dyes. Other
classes of fluorophores which can be used for conjugation chemistry are oxazines
(>600 nm, the spectral shift in absorption and emission is caused by the nitrogen
atom), carbocyanines (~500 to ~700 nm) and coumarins (mostly below ~550 nm).

It is generally approved that fluorescent dyes in the visible spectral region are more
stable, although exhibiting lower quantum yields. Especially under pulsed excitation
conditions at the single molecule level, stability is dramatically increased compared to
short-wavelength fluorophores. This can be explained by the high energy density in
short pulses, which leads to an increase of the probability for two-photon processes.
After two-photon absorption, transitions into higher excited states Sn and Tn can
occur and hence lead to destabilized electronic configurations in molecules, i.e.
increasing the number of antibinding contributions to molecular orbitals [Eggeling et
al, 1998; Eggeling et al., 2005]. In these configurations, photodestruction by light of
short wavelength is more favourable than by light of longer wavelength, e.g. by
photoinduced chemical reactions and generation of a radical ion/electron pair (M+e-).

Besides the advantage of higher photostability observed for chromophores
absorbing in the red spectral region, a lower quantum yield of fluorescence is
generally observed, a result of the increasing influence of nonradiative processes if

                                 Results and Discussion

the energetic gap between HOMO and LUMO becomes smaller. The main process
observed is internal conversion from the lowest vibrational level of S1 to a high
vibrational level of S0, followed by rapid relaxation to the vibrational ground state of
S0. This process is generally attributed to hydrogen stretching vibrations and mainly
expected from hydrogen atoms attached directly to the chromophore [Drexhage

For further examination, three fluorescent dyes from three different classes were
chosen, all absorbing in the red spectral region and available for conjugation
chemistry. Most of them have derivatives over a broad spectral range which makes
them interesting candidates as fluorophores in photonic wires exhibiting an energy
transfer cascade. Molecular structures, spectral properties, fluorescence lifetime and
quantum yields are summarized in figure 4-1.

All three fluorophores are structurally symmetric and only exhibit a small change in
the dipole moment upon excitation from the ground state to the first excited state.
The first fluorophore, Cy5, belongs to the class of symmetric carbocyanines. The
chemical structure contains a polymethine-bridge which connects two indole
moieties. Carbocyanine dyes are generally characterized by high extinction
coefficients, short fluorescence lifetimes and relatively low quantum yield <0.30 (this
is observed if the polymethine backbone is flexible; higher quantum yields are
observed if the structure is rigid, e.g. for Cy3B with a value of 0.65, which is
explained by the influence of structural mobility on nonradiative processes [Drexhage
1973]). Two other fluorophores chosen are the oxazine dye MR121 and the
carborhodamine dye ATTO647. Both exhibit lower extinction coefficients and hence
reduced probability for excitation. Fluorescence lifetimes are considerably longer,
and a high quantum yield of 0.65 is observed for ATTO647.

Time-Resolved Ensemble Spectroscopy

Time-resolved measurements of the three representative dyes mentioned above
were performed using a 635 nm laser diode and a TCSPC based fluorescence
lifetime spectrometer. Dye concentration was chosen to 10-6 M to exclude
interchromophoric interactions or pile-up effects. Since fluorescence properties
substantially depend on the polarity of the solvent, fluorescence lifetimes of
fluorophores were measured in different solvents: phosphate buffered saline (PBS)

                                   Results and Discussion

   Figure 4-1: Three representative dyes absorbing in the red spectral region chosen for
   photophysical experiments. Normalized absorption and emission spectra are shown,
   chemical structure and spectroscopic properties. Extinction coefficient, fluorescence
   quantum yield and lifetime are determined from fluorophores diluted in water.

solution, a 1:1 mixture (v/v) of PBS and acetonitrile, and acetonitrile only. PBS-buffer
can be estimated to have a similar polarity as water (exhibiting a static dielectric
constant of εr = 78 at room temperature) and was chosen since it is commonly used
as standard buffer for biological molecules. Acetonitrile is less polar and exhibits a
static dielectric constant of εr = 37.5 (room temperature).

                                      Results and Discussion

Experimental results of time-resolved ensemble fluorescence measurements are
listed in table 4-1.

                       Solvent                 τ1                  τ2                 χ2

  Cy5                   PBS            0.39 ns (53%)        0.91 ns (47%)            1.04

                PBS:CH3CN 1:1          0.47 ns (48%)        1.17 ns (52%)            1.05

                       CH3CN           0.41 ns (39%)        1.06 ns (61%)            0.93

  ATTO647               PBS            0.92 ns (18%)        3.38 ns (82%)            1.13

                PBS:CH3CN 1:1          1.26 ns (23%)        3.90 ns (77%)            1.11

                       CH3CN           1.76 ns (25%)        4.50 ns (75%)            1.17

  MR121                 PBS            0.45 ns (10%)        1.79 ns (90%)            0.96

                PBS:CH3CN 1:1          0.73 ns (11%)        2.73 ns (89%)            1.00

                       CH3CN           1.64 ns (27%)        4.23 ns (73%)            1.09

Table 4-1: Fluorescence lifetime data derived from time-resolved measurements of the
fluorophores Cy5, ATTO647 and MR121 in solvents of different polarity. Lifetime data was
obtained after applying a biexponential fit, yielding τ1 and τ2 (amplitudes in brackets), and a χ2-
value as a measure of goodness for the fit.

Best fit results on experimental data of fluorescence lifetime were obtained if a
biexponential model was applied. Experiments were carried out with dye samples
obtained directly by the supplier without any further purification. This is likely the main
reason for the inhomogeneity observed in fluorescence lifetimes: one species of a
fluorophore in a homogeneous solvent and constant measurement conditions should
be characterized by one fluorescence lifetime only. Largest inhomogeneities were
observed for the carbocyanine dye Cy5. Nevertheless, the dependency of the
fluorescence lifetime on solvent polarity can be derived from the data shown in table
4-1: the carbocyanine exhibits the smallest changes, and fluorescence lifetime can
be regarded as constant in the range of polarities from the solvents used. The
carborhodamine dye ATTO647 shows an increase of 33%, and the strongest change
is observed for the dye MR121, which more than doubles its fluorescence lifetime in
acetonitrile, compared to aqueous surroundings [Buschmann et al., 2003]. This effect

                                  Results and Discussion

is explained by interactions with the O-H-vibration of water, leading to fast
depopulation of the first excited state and is not observed in D2O [Sauer 1995].

If fluorophores are observed in solvents with higher viscosity or bound to
biomolecules, fluorescence lifetimes are influenced as well. Increasing viscosity
leads to longer reorganization times of solvent molecules surrounding a fluorophore,
hereby affecting the depopulation of the first excited state. In water, reorientation
dynamics of less than 100 ps are observed. Fluorophores bound to biomolecules are
subject to an increased viscosity in the “nano-environment”, yielding similar effects as
in solvents with higher viscosity [Lakowicz, 1995]. A further aspect observed for
many fluorophores leading to an increase in nonradiative depopulation of the excited
state lies in interactions with biomolecules, e.g. quenching by nucleobases or amino
acid residues of a protein. Studies of rhodamines attached to oligonucleotides
revealed    three   different   states   with    different   fluorescent   lifetimes   for
tetramethylrhodamine (TMR) [Eggeling et al., 1998b]: the freely rotating dye exhibits
a lifetime of 2.2 ns (which is similar to the fluorescence lifetime of the free dye), the
dye closely bound to nucleobases (except guanosine) shows a broad distribution
around 3.5 ns as a result of a more immobile molecule, and a third lifetime of 0.8 ns
together with a strongly decreased quantum yield is observed if interactions with
guanosine residues occur due to electron transfer from guanosine to the dye
[Knemeyer et al., 2000].

In the following passages, single-molecule experiments with these fluorophores using
the confocal set-up developed in the course of this work (see 3.1) will be discussed.
To circumvent the inhomogeneities and broadening of fluorescence properties known
from experiments with fluorophores adsorbed randomly on a glass substrate [Weston
et al., 1998] experiments were carried out in liquid environment. Single-molecule
measurements were realized by conjugating the fluorophores to the 5’-end of a
double-stranded 60 base pairs DNA with a biotin anchor on its 3’-end. As described
in the experimental section (see 3.3.2), such fluorophore-DNA-conjugates can be
immobilized onto a protein-surface and measured in liquid environment [Anzai et al.,
1995; Piestert et al., 2003]. The preparation technique allowed the elaboration of the
influence on two main effects onto the photophysical behaviour of fluorophores, i.e.
the presence of oxygen and the presence of electron-donating substances. Oxygen
was removed enzymatically from the buffer solution [Funatsu et al., 1995]. To

                                  Results and Discussion

prepare a reductive environment, a solution of 100 mM mercaptoethylamine (MEA) in
PBS was used (described in section 3.3.3).

Experiments in the following section were made in the following chemical

                   PBS only (used as reference)

                   PBS with 100 mM MEA, i.e. reductive environment

                   PBS with oxygen removed enzymatically

                   PBS with 100 mM MEA and oxygen removed.

Single dye molecules were excited with a helium-neon laser with a centre wavelength
of 632.8 nm. Fluorescence was recorded on two spectrally separated detectors by
using a beamsplitter with a cutting edge at 680 nm (680DRLP; Omega Filters, USA)
and two bandpass filters (675DF50, 700DF75; Omega Filters, USA). Fluorescence
scan images were generally recorded with a spatial resolution of 50 nm per pixel and
an integration time of 2 ms per pixel. To derive kinetic information from photon
statistics of single molecule fluorescence trajectories, both autocorrelation and
duration histogram analysis were carried out (these methods are described in 2.2.3)
by   using custom-written software (LabView 7, National Instruments, USA). Time
resolution of autocorrelation analysis was set to 1 µs (this time resolution was chosen
due to practical reasons; the limit is set by the time resolution of the macroscopic
time of FIFO data, i.e. 50 ns, but leads to a dramatic increase in calculation time and
only poor additional information in single-molecule experiments, as the total number
of photons detected is limited). Duration histograms were only used when
fluorescence fluctuations were accessible by this method, i.e. a clear distinction
between “on” and “off” states was possible. Fluorescence trajectories were binned in
10 ms intervals (if not stated otherwise).

                                 Results and Discussion


The carbocyanine dye Cy5 is the red representative of a series of symmetric
polymethine dyes with indole moieties. Particularly in single-molecule experiments,
this class of fluorophores has been used for many applications, from mere
fluorescence detection if bound to biomolecules over FRET-based experiments to
complex trichromophoric assays [Yildiz et al., 2003; Diez et al., 2004; Hohng et al.,
2004]. Another aspect intensively investigated for Cy5 are complex photophysical
reactions, such as isomerisation from a fluorescent trans-configuration to a
nonfluorescent cis-configuration together with intersystem crossing into triplet-states
were investigated [Widengren and Schwille, 2000]. It was observed that the
isomerisation process can lead to delayed fluorescence involving a thermal-induced
backisomerisation out of the cis-triplet state, occurring in the timescale of
phosphorescence [Huang et al, 2005]. The photophysical pathway towards
irreversible photodestruction was found to be composed of several steps with
different dependencies on excitation intensity or the presence of oxygen [Ha and Xu,
2003; Füreder-Kitzmüller et al., 2005]. Very recently, it could be demonstrated that
Cy5 fluorophores in solution could be reactivated to fluorescence emission out of a
nonfluorescent dark-state by an electric field [White et al., 2004]. Even more recently,
new studies and part of the present work demonstrated that all-optical switching of
Cy5 molecules by applying a second laser wavelength is possible and can be
repeated up to 100 times [Heilemann et al., 2005; Bates et al., 2005]. In this section,
fluorescence fluctuations of individual Cy5 molecules bound to DNA and observed in
different aqueous environment will be the focus. Fluctuations are analysed by using
the autocorrelation function and duration histograms. Optical switching of single
carbocyanine molecules will be discussed in more detail in a section later.

Fluorescence scan images, trajectories and autocorrelation curves of individual Cy5-
molecules recorded in different chemical environments are presented in figure 4-2
(Cy5 in PBS with 100 mM MEA is omitted at this point, since no influence on
fluorescence properties is observed). The scan image (A) recorded in PBS reflects
the poor photostability of carbocyanine dyes generally observed. From 14 molecules
observed in the area of 5 x 5 µm², 8 molecules are photobleached during scanning

                                    Results and Discussion

Figure 4-2: Fluorescence scan images of Cy5 in (A) PBS, (B) PBS with oxygen removed and
(C) PBS/MEA with oxygen removed (50 nm pixel, integration time / pixel = 2 ms, 2 kW/cm²,
intensity scale 10-40 counts / 2 ms). In similar environment, fluorescence trajectories (D)-(F)
were recorded and autocorrelation functions of trajectories were calculated, (G)-(I).

the image, representing a fraction of more than the half the molecules under
moderate excitation conditions (2 kW/cm²). Upon removal of oxygen from the
solution, long “off”-periods appear in image (B). These “off”-states are generally
attributed to triplet-states which are efficiently quenched in the presence of
molecular-oxygen [Veerman et al., 1999, Tinnefeld et al., 2001] or charge separated
states formed via the triplet state [Zondervan et al., 2003]. Although exhibiting severe
intensity fluctuations, fluorescence spots of molecules show significantly less

                                   Results and Discussion

photobleaching than observed for experiments made in PBS. Fluorescence blinking
of Cy5 molecules is strongly reduced in PBS/MEA, portrayed in scan image (C). In
the case of carbocyanines, MEA acts as efficient triplet quencher in the absence of
oxygen [Widengren and Schwille, 2000]. Compared to the scan image recorded in
PBS, a considerably lower number of fluorescence spots exhibits photobleaching.

Fluorescence trajectories (D)-(E) in figure 4-2 demonstrate the influence of the
chemical environment on fluorescence intensity pattern. In PBS, photobleaching is a
rapidly occurring event, and mainly short trajectories are observed. A mean value of
3400 photons was determined from 22 individual Cy5 molecules in PBS. Oxygen
removal leads to “off” periods on the millisecond timescale, which again are reduced
dramatically by adding MEA. Both environments lead to drastically longer
observation times: a mean number of 85000 photons (oxygen-free PBS) and 45000
(PBS/MEA) was determined from 39 and 14 individual trajectories, respectively.

Fluorescence trajectories were further analysed by correlating fluorescence
fluctuations, yielding autocorrelation functions (G)-(I) (grey curves). Results of multi-
exponential fits (red curves) are listed in table 4-3.

                         τ1,on / ms τ1,off / ms τ2,on / ms τ2,off / ms τ3,on / ms τ3,off / ms

 PBS                       0.015      0.010         40         2.8          -          -

 PBS, oxygen removed       0.090      0.076         14         2.7        120         23

 PBS 100 mM MEA            0.115      0.053         150        11           -          -
 oxygen removed

Table 4-3: Correlation times derived from approximation of autocorrelation curves by
exponential fit.

For Cy5 in PBS, two correlation times are observed. According to results published in
literature, the shorter contribution is attributed to cis-trans isomerisation of the
polymethine backbone, and the longer contribution to intersystem crossing to the
triplet state [Tinnefeld et al., 2001; Widengren and Schwille, 2000; Heilemann et al.,
2005]. In oxygen-free environment, two longer contributions to the correlation
function appear and are more dominant than in the presence of oxygen. Here, the
absence of oxygen leads to the observation of longer triplet states now expanding to

                                 Results and Discussion

milliseconds. If MEA is added to the solution, correlation times observed lack the long
contribution observed before, and exhibit two components with a comparable
timescale to those observed in PBS.

If oxygen is removed, an increase in nonfluorescent periods is observed which are
attributed to longer triplet-states, formerly quenched in the presence of oxygen. As a
second effect, the observation time prior to irreversible photodestruction increases.
Therefore, it is generally assumed that molecular oxygen – after depopulating the
triplet state of the carbocyanine and forming very reactive singlet oxygen – plays an
important role in the photobleaching pathway of Cy5 [Ha and Xu, 2003]. This is
supported by experiments with MEA as triplet quencher for carbocyanines where
shorter characteristic times appear in the autocorrelation.

Interestingly, other reductive agents containing a thiol-moiety induce a similar effect
as observed for MEA. Here to mention are β-mercaptoethanol (BME), cystein or
dithiothreitol (DTT). In a general view, these reagents have electron-donating
properties and create a reductive environment. In the presence of carbocyanines,
they demonstrate important triplet depopulation of the fluorophores [Widengren and
Schwille, 2000].

Fluctuations in fluorescence intensity observed for the carbocyanine dye Cy5 are in
agreement with published data. Furthermore, it could be demonstrated that
carbocyanines can be stabilized if molecular oxygen is removed, allowing a 25-fold
increase in the total number of photons detected and hereby dramatically longer
obervation times of single fluorophores. In the following sections, similar chemical
environments are applied in experiments with other fluorophores, which belong to the
classes of oxazines and rhodamines.

                                 Results and Discussion


The fluorescent dye MR121 is the representative chosen out of the structural class of
oxazine dyes (available as DYE121, AttoTec, Germany; single-molecule experiments
discussed in this section were carried out with a structural very similar dye,
ATTO655, which shows nearly identical photophysical properties but is better soluble
in aqueous media and more suitable for conjugation to DNA). This class of dyes is
frequently used in experiments exploiting PET for small structural changes, since
fluorescence emission is efficiently quenched in the presence of tryptophane or
guanosine [Neuweiler et al., 2002; Marmé at al., 2003; Doose et al., 2005; Heinlein et
al., 2003]. In this context, the focus was set on fluorescence intermittencies of oan
oxazine dye appearing in different chemical environment (figure 4-3), in comparison
to carbocyanines and rhodamines.

      Figure 4-3: Fluorescence trajectories of ATTO655 bound to DNA in (A) PBS, (B)
      PBS with 100 mM MEA, (C) oxygen-free PBS and (D) oxygen-free PBS with 100
      mM MEA (2 kW/cm², 10 ms binning time).

                                   Results and Discussion

Oxazine dyes already exhibit constant and bright fluorescence emission in PBS. At
the timescale of 10 ms, almost no fluorescence intermittencies are observed (see
trajectory (A) in figure 4-3). This situation changes in reductive environment:
fluorescence “spikes” are observed and long intermittencies on timescales of up to
hundreds of milliseconds appear (trajectory (B) in figure 4-3). If oxygen is removed,
nonfluorescent periods increase in length and “on”-periods get very rare, further
reduced in intensity if MEA is added (trajectories (C) and (D) in figure 4-3). As a first
result, it is evident that ATTO655 is suited for measurements in PBS only, but is not
suitable for experiments in oxygen-free or reductive environment. Scan images
shown in figure 4-4 further emphasize that the behaviour exemplified on individual
molecules represents indeed a general characteristic of ATTO655 trajectories in the

 Figure 4-4: 10 x 10 µm² scan images of ATTO655 bound to dsDNA in (A) PBS, (B) PBS with
 100 mM MEA, (C) PBS oxygen removed, (D) PBS with oxygen removed and 100 mM MEA
 (Excitation power 2 kW/cm², 50 nm pixel size and 2 ms integration time / pixel; intensity
 scale from 15-40 counts / 2ms).

respective environment. The scan images recorded in PBS show brightest
fluorescence emission and nearly no fluorescence intermittencies (image (A) in figure
4-4). Pronounced fluctuations are observed in PBS/MEA (image (B)) and reproduce
“off”-periods of several milliseconds, according to observations made in fluorescence
trajectories. A considerably lower density of molecules due to long “off”-periods is
observed if oxygen is removed from the solution (image (C) and (D)).

Autocorrelation curves and exponential fits of fluorescence trajectories in PBS and
PBS/MEA are presented in figure 4-5. No significant correlation term down to 1 µs is
observed for ATTO655 in PBS. This explains the use of this class of fluorophores for
fluctuation analysis in peptides, DNA or RNA molecules, since they do not exhibit any
triplet transitions or other transitions into nonfluorescent states [Doose et al., 2005]. If

                                   Results and Discussion

 Figure 4-5: Autocorrelation analysis of fluorescence trajectories from ATTO655 in (A) PBS
 and (B) PBS with 100 mM MEA (binning width 1 µs). The red curve in (B) represents
 monoexponential approximation of autocorrelation data.

MEA is added, the autocorrelation function exhibits fluctuations in the timescale of a
few milliseconds, and a monoexponential fit function yields an “on”-time of 33.2 ms
and an “off”-time of 16.6 ms. The longer “off”-periods which are clearly visible in
trajectory (B) of figure 4-3 can not be identified in the autocorrelation function, but
can be derived from duration histogram analysis, yielding a value of 315 ms (the
“on”-time derived from duration histograms has a value of 30 ms and therefore is
identical to the one derived from autocorrelation analysis). The analysis of 49
fluorescence trajectories from individual ATTO655 molecules delivered mean values
for fluorescence intermittencies listed in table 4-4, obtained by both autocorrelation
and duration histograms.

                                                  τon / ms       τoff / ms

             Autocorrelation                      29 (20)         19 (14)

             Duration Histograms                  19 (8)        378 (124)

Table 4-4: Mean values for “on”- and “off”-periods calculated from 49 single ATTO655
fluorophores in PBS/MEA, standard deviations in brackets.

The results from photon statistics listed in table 4-4 suggest the presence of two “off”-
states, a short one with 19 ms and a longer one in the order of 378 ms. Both “off”-
times are not observed in the absence of MEA, and are therefore assigned to

                                   Results and Discussion

electron transfer towards the fluorophore in reductive environment. The broadness of
the distributions suggests a certain heterogeneity which influences the back-
transition from this dark-state to the ground state or an excited state.

Fluorescence trajectories in the absence of oxygen are more difficult to evaluate.
Several characteristic terms on largely different timescales are observed in
autocorrelations, and did not allow unambiguous analysis of photon statistics.
Fluorescence intermittencies under these experimental conditions are usually
ascribed to triplet transitions, but usually exhibit similar signatures for different
fluorophores of one kind.

It can be presumed that MEA as reductive agent is acting similar to tryptophane and
guanosine, i.e. electron transfer to the fluorophore and hereby creating a radical
anion [Doose et al., 2005]. In this state, the fluorophores do not exhibit fluorescence
anymore. This radical anion exhibits a certain lifetime which strongly depends on its
chemical environment, i.e. the ability of the solvent to stabilize the radical state.

The mechanism observed is best characterized by PET and exploited for short-range
changes in conformation of biomolecules [Neuweiler et al., 2002; Marmé et al.,
2003]. In the context of this work, the use of oxazine dyes is problematic, because
their redox potential easily accesses these fluorophores for electron transfer into a
nonfluorescent state.

                                 Results and Discussion


The carborhodamine dye ATTO647 is the last dye of the representative choice of
chromophores. Structurally, a carborhodamine is closely related to rhodamines: the
central oxygen-atom is replaced by a methylen group.

Exemplary fluorescence trajectories in four respective environments are shown in
figure 4-6. As visible from fluorescence trajectory (A), the fluorophore exhibits long
fluorescence intermittencies in the range of hundreds of milliseconds, not observed
for either oxazines or carbocyanines. This typical behaviour for ATTO647 bound to
DNA and in aqueous solution is observed for all molecules, but on strongly different
timescales. In reductive environment, no fluorescence intermittencies are observed
anymore, as shown in trajectory (B). This behaviour resembles the one observed for
quantum dots, which do exhibit intensity fluctuations at different timescales which can

      Figure 4-6: Fluorescence trajectories of ATTO647 bound to DNA in (A) PBS, (B)
      PBS with 100 mM MEA, (C) oxygen-free PBS and (D) oxygen-free PBS with 100
      mM MEA (3 kW/cm², 10 ms binning time).

                                        Results and Discussion

be suppressed by electron-donating reagents, e.g. MEA or BME [Hohng and Ha,
2004]. Total photons detected are ~260000 in (A) and ~400000 in (B). This reflects a
tendency to higher photon counts detected from single fluorophores observed for
most trajectories: mean values of 163000 and 270000 photons detected were
determined from 11 and 22 single-molecule trajectories, respectively. Remarkably,
some trajectories reach a total photon detection of nearly 106 per molecule in
reductive environment. The observation time until photodestruction of single
ATTO647 molecules under both conditions is very similar. Therefore a comparable
number of excitation cycles probably occur, but with the difference that a large
number of these cycles do not lead to fluorescence in pure buffer.

Upon oxygen-removal, fluorescence trajectories are characterized by “spikes” of
fluorescence emission (part (C) of figure 4-6), comparable to carbocyanine dyes.
Contrary to Cy5, these intermittencies are not suppressed by adding MEA (part (D)).
To obtain a more qualitative description of fluorescence kinetics observed,

    Figure 4-7: Autocorrelation functions (black) and exponential fit (red) derived for
    fluorescence trajectories of ATTO647 in (A) PBS, (B) PBS/MEA, (C) oxygen-free PBS
    and (D) oxygen-free PBS/MEA; mono-exponential fits in (A)-(C) and a three-
    exponential fit in (D) are shown.

                                    Results and Discussion

fluorescence fluctuations were evaluated applying the autocorrelation function.
Curves calculated for the fluorescence trajectories presented above are depicted in
figure 4-7.

Correlation curves (A)-(C) were approximated by a monoexponential fit function,
whereas a three-exponential fit was applied for correlation curve (D). Characteristic
times derived from autocorrelation analysis are listed in table 4-6.

                                            τchar / ms        A       τon / ms τoff / ms

   PBS                                         86.7          0.87       186       162

   PBS 100 mM MEA                             0.038          0.045      0.88     0.04

   PBS Oxygen-free                             0.36          8.88       0.40     3.56

   PBS Oxygen-free 100 mM MEA                 0.087          0.68       0.21     0.15

                                               1.03          0.75       2.4       1.8

                                               30.3          0.56           84    47

Table 4-6: Characteristic times for “on” and “off” states observed in ATTO647 fluorescence
trajectories, derived from autocorrelation and exponential approximation.

Fluorescence fluctuations shown in figure 4-7 merit a further discussion. In reductive
environment (B), a short time of 40 µs appears which does not in (A). This leaves the
question if this component was neglected due to a ~20fold lower amplitude, or if it
arose in the presence of MEA. Concerning the second case, this might be explained
by reduction of the fluorescent state of ATTO647 to a radical anion (comparable to
ATTO655). Another possible interpretation starts with the fact that “off” times
observed in PBS and oxygen-free PBS differ strongly, yielding 162 ms and 3.56 ms
respectively. This suggests that species of different nature are generating this
nonfluorescent state, which are created by the nanoenvironment. In the presence of
oxygen (A), the photooxidized species of ATTO647 is observed. In the presence of
MEA (B), this species is vanishing, and a short-lived species appears which could
represent the reduced state. In the absence of oxygen (C), a shorter time could
represent the triplet state of ATTO647. The correlation curve in (D) shows a longer
and shorter time in the millisecond range, but fluorescence intermittencies observed

                                           Results and Discussion

under these conditions seem to exhibit “off”-times on many timescales and cannot be
characterized properly. Derived from a number of different correlation curves under
equal conditions, different characteristic times were determined, which supports the
fact that changes of transition rates with time are observed.

Characteristic “on”- and “off”-times of ATTO647 in PBS lie in the 100 ms-range and
can be determined as well from duration histograms. Applying this method for
trajectory (A) in figure 4-6, values of 188 ms and 171 ms for the “on”-state and the
“off” state are obtained and are in good agreement with values derived from
autocorrelation (see table 4-6).

It   has   to     be     mentioned         that
characteristic times for “on” and “off”
states          vary          substantially.
Fluorescence trajectories recorded in
PBS exhibit a broad distribution for
both times, which suggests a great
heterogeneity that may arise from
interactions      with      the       scaffold
molecule, DNA. Values obtained by
                                                   Figure 4-8: Distribution of “on“- and “off“-times
autocorrelation        analysis       of    40
                                                   derived   from    40   individual   fluorescence
fluorescence           trajectories         are
                                                   trajectories of single ATTO647 molecules.
presented in figure 4-8. A mean
value of 211 ms is obtained for collected “on” times, and 341 ms for “off”-times
(standard deviation of 122 ms and 161 ms respectively). No obvious correlation
between “on” and “off” states could be observed. A similar inhomogeneous behaviour
for “on”- and “off”-periods is observed for trajectories recorded in other chemical

To summarize, it can be stated that constant fluorescence emission without
transitions to intermediate states with nonfluorescent nature are observed in
reductive environment, created by 100 mM MEA. Same results were obtained when
MEA was replaced by BME, which supports the thesis of a dominating influence of
the reductive potential of the SH-group present in both substances. In a more
depictive way, the fluorescence intermittencies of ATTO647 bound to DNA can nicely
be seen on fluorescence scan images, shown in figure 4-9.

                                      Results and Discussion

 Figure 4-9: 10 x 10 µm² scan images of ATTO647 bound to dsDNA in (A) PBS, (B) PBS with
 100 mM MEA, (C) PBS oxygen removed, (D) PBS with oxygen removed and 100 mM MEA
 (Excitation power 2 kW/cm², 50 nm pixel size and 2 ms integration time / pixel; intensity
 scale from 20 – 100 counts / 2ms).


Promising results obtained for the carborhodamine dye ATTO647 in the previous
sections were subject for further investigation on the class of rhodamine dyes, which
is structurally closely related to carborhodamines. They offer the biggest group of
fluorophores which is commercially available. In table 4-7, a list of three rhodamine
dyes and their spectroscopic properties in water are listed.

                              λabs               λem            τfl              φfl

Rhodamine Green               508                534           3.70             0.9

TMR                           560                582           2.20             0.9

ATTO590                       603                625           3.70             0.8

Table 4-7: Spectroscopic properties of rhodamine dyes RhG, TMR and ATTO590 bound to
dsDNA in water. Fluorescence lifetimes represent the population of fluorophores that freely

Single-molecule fluorescence studies of rhodamine derivatives are the topic of the
following sections. Different to previous experiments, a continuous-wave argon laser
operating at 488 nm was used as excitation source. Excitation energies were

                                  Results and Discussion

~5 kW/cm². Fluorescence was recorded with up to four detector channels using the
standard filter set described in materials and methods (see 3.1).

In the first section, single-molecule fluorescence studies of Rhodamine Green are
presented. The fluorophore was used as input unit for photonic wire constructs (see
4.2 and 4.3). In the following sections, FRET-pairs constructed exclusively of
rhodamine derivatives will be discussed.

Single-Molecule Experiments with RhG in PBS and Reductive Environment

Fluorescence trajectories of single Rhodamine Green molecules are shown in figure
4-10 (A) and (B). Typically, trajectories show a constant emission of fluorescence
without fluorescence intermittencies at a timescale of the binning chosen (10 ms). A
background fluorescence intensity of below 1 kHz is observed.

Fluorescence trajectories in reductive environment are presented in (C) and (D) of
figure 4-10 and show a significantly increased photostability. On the other hand, the
count rate is not affected. Trajectories (C) and (D) exhibit a survival time of ~20 s and
~15 s before irreversible photobleaching. As a consequence of the changed chemical
environment, higher background intensities are now observed, reaching up to 3 kHz
as a sum of both detectors and hereby representing a threefold increase compared
to measurements in PBS only. This observation was not made when using red laser

To quantify the increased number of photons emitted by single Rhodamine Green
fluorophores in reductive environment, statistic evaluations of the total number of
photons before irreversible photobleaching were carried out. Histograms obtained
from ~80 fluorescence trajectories measured in PBS and ~40 trajectories measured
in PBS with 100 mM MEA are shown in part (E) of figure 4-10. The blue bars
represent the number of emitted photons of fluorophores collected in PBS, showing a
distribution at lower photon counts with a mean value of 24000 photons/molecule
(this value is comparable to what was found for Rhodamine6G in water, with on
average 25000 photons per molecule [Mets and Rigler, 1994]). The red bars
represent the number of photons emitted in PBS with MEA, and show a distribution
with two distinguishable “populations”: at lower photon counts, the distribution

                                    Results and Discussion

   Figure 4-10: (A) and (B): RhG fluorescence trajectories of single fluorophores recorded
   in PBS; (C) and (D): fluorescence trajectories recorded in PBS with 100 mM MEA; (E):
   Total photon counts until photobleaching of single fluorophores occurred (blue bars for
   RhG in PBS, red with 100 mM MEA added); (F): Autocorrelation functions of single
   tracjectories (1 µs binning) do not show significant fluctuations.

resembles the one derived from measurements in PBS. Additionally, a second
distribution around 200000-250000 photons/molecule can be observed.

Strikingly, fluorescence trajectories observed for Rhodamine Green in reductive
environment resemble those of the carborhodamine ATTO647 (see 4.1.4). For both
fluorophores, a similar number of photons can be detected prior to photobleaching.

                                    Results and Discussion

Different between both fluorophores is the fluorescence signature observed in pure
buffer. This could suggest the formation of an oxidized species in both cases, with
the difference that the transition towards a possible radical cation is reversible for
ATTO647 and irreversible for Rhodamine Green.

To reveal fluorescence intermittencies at a lower timescale as derived from
trajectories with a binning of 10 ms, autocorrelation functions with a time resolution of
1 µs were calculated, see part (F) in figure 4-10. Due to a better signal-to-noise ratio,
autocorrelations were calculated from fluorescence data detected on the blue
channel. For both PBS (blue curve) and PBS/MEA (red curve), no significant intensity
fluctuation can be observed down to a correlation time of 10 µs (which is in
agreement with triplet times below 1 µs and poor triplet quantum yields observed for
rhodamine dyes in water with low excitation energies, [Widengren and Rigler, 1996]).
At shorter observation times, a steep rise in the correlation function appears which is
sometimes obtained with such low photon statistics and is attributed to afterpulses
from the detectors (APD) [Zhao et al., 2003].

Experiments in Oxygen-free Environment

Fluorescence trajectories of Rhodamine Green in PBS with oxygen removed are
shown in part (A) of figure 4-11. Here, two interesting differences to fluorescence
trajectories observed for carbocyanines do appear: first, the total measurement time
is in the same order of magnitude as for measurements in PBS containing oxygen.
Contrary to Cy5 in oxygen-free solution, this means that molecular oxygen does not
participate dominantly in the photobleaching pathway of Rhodamine Green. On the
other hand, fluorescence intermittencies are now observed, and well-distinguishable
levels of “on” and “off” appear in single-molecule trajectories. Similar fluctuations are
observed in reductive and oxygen-free environment, shown in an exemplary
trajectory in part (B) of figure 4-11.

Additionally, a similar effect on fluorescence trajectories is observed as presented in
figure 4-10 for Rhodamine Green in PBS/MEA, i.e. a considerably increase in the
total lifetime of the fluorophore before irreversible destruction. Total photon counts for
both experimental conditions of ~40 molecules are histogrammed in part (E) of figure
4-11. For oxygen-free PBS only (blue bars), a narrow distribution with lower photon
counts is observed. A mean value of 9500 counts/molecule is calculated, which

                                     Results and Discussion

 Figure 4-11: RhG fluorescence trajectories of a single fluorophore recorded in oxygen-free
 PBS, (B) 100 mM MEA added. Autocorrelation curves of the same fluorophores are shown
 in oxygen-free PBS (C) and oxygen-free PBS/MEA (D). Statistical data on total photon
 counts is shown in (E), autocorrelation fit values and “on”- and “off”-times are listed in (F).

represents less than 40% from total photon counts if oxygen is present. In reductive
and oxygen-free solution (red bars), a broad distribution with a mean value of
~200000 photons/molecule is determined, which is comparable to the oxygen-
containing PBS/MEA measurements.

In addition, fluctuations of single-molecule fluorescence trajectories in oxygen-free
solution were analysed by autocorrelation, as shown in part (C) (oxygen-free PBS)
and part (D) (oxygen-free PBS/MEA) of figure 4-11. In both environments, correlation

                                    Results and Discussion

times by applying a mono- or biexponential model can be derived. Low photon
counts in oxygen-free environment led to poor statistics in the autocorrelation
function, and a reasonable approximation could only be achieved applying a
biexponential model. The first time constant is not taken into account due to
important interference with afterpulsing characteristics from the APDs. The second
correlation time of 9.85 ms is comparable to the one derived from oxygen-free
PBS/MEA measurements, 7.84 ms. Both correlation functions lead to similar values
for the “on” state and the “off” state (figure 4-11, (F)).

Fluorescence intermittencies of 15 (oxygen-free) and 13 (oxygen-free/MEA) single-
molecule trajectories were analysed by applying the autocorrelation function and
duration histograms. Mean values for “on” and “off” states derived from these
calculations together with standard derivations are presented in table 4-8:

                         Autocorrelation Analysis               Duration Histograms

                       τon / ms           τoff / ms          τon / ms          τoff / ms

O2-free                51 (28)            19 (9.0)           36 (12)           40 (15)

O2-free/MEA            51 (25)            8.7 (1.3)          56 (16)            20 (2)

Table 4-8: Photon statistics of “on“- and “off“-states from single Rhodamine Green molecules
in oxygen-free PBS in the presence and absence of MEA, derived by autocorrelation or
duration histogram analysis.

Because of relatively low fluorescence count rates observed for single Rhodamine
Green trajectories, trajectories had to be binned with 10 ms to sufficiently distinguish
“on”- from “off”-states. As a consequence, duration histograms are less suited to
derive characteristic times around 10 ms than autocorrelation analysis is. This
explains the deviation of the “off”-state lifetime from the value determined by applying
autocorrelation analysis. Smaller values for the standard deviations are obtained for
oxygen-free/MEA trajectories, which is explained by drastically longer observation
times. To summarize, an “off”-state with a lifetime around 19 ms for oxygen-free and
8.7 ms for oxygen-free/MEA environment suggests the presence of a relatively long-
lived species which is suppressed if oxygen is present. It can be anticipated that this
species is a triplet state of Rhodamine Green, which is effectively quenched by

                                   Results and Discussion

oxygen and appears to interact slightly with MEA. The quenching of the triplet-state
by MEA is substantially smaller than observed for carbocyanines.

As already mentioned in the discussion of the autocorrelation function shown in (D)
of figure 4-11, short times appear for oxygen-free/MEA environment, but may suffer
from overlap with afterpulsing from APDs. No such short time is observed in
experiments in oxygen-containing environment. Autocorrelation analysis with a
biexponential model were applied to 12 appropriate fluorescence trajectories which
yielded a satisfactory quality of the fit-function. Values obtained for “on”- and “off”-
times are

                τon = 0.17 ms (0.14 ms) and

                τoff = 0.06 ms (0.05 ms).

It is difficult to further characterize the nature of this state. Short contributions to the
autocorrelation function implicate large errors and distributions are broad. These
contributions are only observed if oxygen is removed and no MEA is added, a similar
observation was made for carbocyanine derivatives and is attributed to the
appearance of long-lived triplet states (see 4.1.2). Low photon statistics and short
fluorescence trajectories do not allow any further examination of these states.

                                    Results and Discussion


Single-molecule FRET (smFRET) is a widely used tool to investigate distance
fluctuations at the single-molecule level in the range of 2 to 10 nm [Ha et al., 1996].
Fluorescence intensity of both a donor and an acceptor molecule are monitored and
fluctuations in energy transfer derived from the emission spectrum of the FRET-pair
are attributed to distance changes in the system under investigation. Therefore, it is
desired to use a pair of fluorophores where each component does not exhibit any
intrinsic intensity fluctuations, leading to a misinterpretation of experimental data [Ha
et al., 1999].

Results obtained for Rhodamine Green and ATTO647 suggest the use of
rhodamines in single-molecule FRET-applications. As the last section discussed in
detail (see 4.1.5), measurement conditions best suited for this class of chromophores
is an oxygen-containing and reductive environment, e.g. 100 mM MEA. In a next
step, a double-stranded DNA was constructed with Rhodamine Green at the 5’-end
of one single strand and TMR spaced by 10 base pairs at the opposite single strand.
In this configuration, the fluorophores are spaced by ~3.4 nm, and the rigid structure
of DNA should not reveal structural dynamics. The R0-value calculated for this FRET-
pair equals 6.5 nm, so high transfer efficiency can be expected. Fluorescence
emission was recorded on the three detector channels denoted “blue”, “green” and
“yellow” (see section 3.1).

Fluorescence emission of Rhodamine Green is mainly recorded on the blue and
green detector. The spectral distribution on different detectors can be described more
pictorial by introducing a fractional intensity, F, defined as the fraction of intensity
recorded on the long-wavelength detector divided by the sum of intensities,

                                                    I Det 2
                              FDet 2, Det1 =                                     ( 4-1 )
                                               I Det1 + I Det 2

Here, IDet2 stands for the photons recorded on the long-wavelength detector, and IDet1
for photons recorded on the short-wavelength detector. For Rhodamine Green only,
a fractional intensity of Fgreen,blue = 0.30 is calculated from single-molecule
fluorescence trajectories, and fluorescence detection on the “yellow” channel can be

                                  Results and Discussion

neglected. For the acceptor TMR, practically no emission is observed on the “blue”
channel, and a value of Fyellow,green = 0.42 was determined (via direct excitation).

Fluorescence trajectories of the sm-FRET-pair RhG-TMR in PBS and PBS/MEA in
oxygen-containing environment are shown in figure 4-12. Trajectories recorded in
PBS suffer from short measurement times and rapid photodestruction (part (A)),
whereas the reductive environment induced by 100 mM MEA leads to longer
observation times and a similar increase in photon counts as observed for
Rhodamine Green only. Fractional intensity traces and histograms exhibit resembling
characteristics: calculated values from photons detected on the “blue” and “green”
channel show a distribution with two populations with maximum values of Fgreen,blue =
0.36 and 0.68 (PBS, (B) in figure 4-12) and of Fgreen,blue = 0.27 and 0.70 (PBS/MEA,
(C) in figure 4-12). The first value can be assigned to the part of the traces where
Rhodamine Green is emitting predominantly, that is, after photodestruction of the
acceptor chromophore TMR. The second value stems from the first section of the
fluorescence trajectories exhibiting energy transfer and is a measure of the efficiency
of this process. In an ideal case with an energy transfer efficiency approaching 100%
and a low fraction of background-to-signal, this value should be close to 1, since the
emission spectrum of TMR does not show any significant contribution on the “blue”
detector, i.e. below 548 nm. From both trajectories, a value of around 0.70 is derived
and denotes efficient energy transfer with dominant emission from TMR. For the
fractional intensity calculated from photon detection on the “green” and “yellow”
channel, a mean value of Fyellow,green = 0.37 (PBS) and Fyellow,green = 0.35 was
calculated. Due to similar reasons discussed before, this value is slightly lower as
expected from ensemble spectral data for TMR, but suggests high energy transfer

Both traces show a relatively high efficiency of energy transfer. An absolute value
can be derived from the fluorescence emission of Rhodamine Green detected on the
“blue” channel. Assuming no relevant contribution from TMR on the “blue” channel
and a constant value of Fgreen,blue for Rhodamine Green, the intensities recorded on
this detector during the presence of the acceptor and after the photodestruction of
the acceptor can be used for calculation of transfer efficiencies. Values obtained are
0.86 (PBS) and 0.75 (PBS/MEA), which confirms the highly efficient transfer of
excited state energy expected in this construct. Interestingly, neither of the
trajectories exhibits fluctuations in energy transfer. Trajectory (D) in figure 4-12
                                      Results and Discussion

Figure 4-12: (A): Spectrally-resolved single-molecule trajectory of the FRET-pair RhG-(N)10-
TMR measured in PBS, with fractional intensities Fgreen,blue (green) and Fyellow,green (yellow),
histogrammed in (B) and (C). Longer observation times are observed in PBS/MEA (D) with
similar spectral distribution, (E) and (F).

shows another photophysical phenomenon, which is a collective “off” state of both
donor and acceptor, which could be the result of a radical state of RhG or TMR.

As a logical consequence, one might now think of adding more fluorescent dyes and
further extend the system. This was realized by “adding” two more fluorophores,
ATTO590 and ATTO647, with identical spacing of 10 base pairs or 3.4 nm.
Fluorescence trajectories recorded in the presence of 100 mM MEA are shown in
figure 4-13.

Trajectories (A) and (B) show spectrally resolved fluorescence intensity emitted by a
construct of three fluorophores: RhG, TMR and ATTO590. A fourth fluorophore was
added in trajectories (C) and (D), i.e. ATTO647. A fourth detector channel was also

                                     Results and Discussion

 Figure Figure 4-13: Single-molecule fluorescence trajectories of a three-dye FRET-pair RhG-
 (N)10-TMR-(N)10-ATTO590 in (A) and (B) and a four-dye FRET-pair RhG-(N)10-TMR-(N)10-
 ATTO590-(N)10-ATTO647 in (C) and (D). Trajectories are recorded in PBS/MEA and recorded
 on four spectrally-separated detectors to better reflect single steps of energy transfer.

added, denoted “red” and recording photons in a spectral region from 680 to 738 nm.
All trajectories exhibit long observation times, showing the applicability of stabilizing
rhodamine-like fluorophores in reductive environment. However, three and four
fluorophores already present complicated multichromophoric compounds and exhibit
complex photophysical behaviour, which shall not be discussed any further in this
section, but will be subject of an own section later in this work (see 4.2 and 4.3).

At this point, it could be demonstrated that rhodamine and carborhodamine dyes can
be stabilized in aqueous solution to a factor of 10, regarding total photon counts.
Precisely, this was demonstrated for the fluorophores Rhodamine Green, TMR,
ATTO590 and ATTO647. For the carborhodamine dye ATTO647, a complete
suppression of fluorescence intermittencies could be realized in the same
environment. Furthermore, smFRET-pairs of up to four fluorophores could be shown
                                 Results and Discussion

to be substantially more stable, allowing for longer observation times. It can be
anticipated that these results are assignable to fluorophores with a similar chemical
structure, i.e. many other fluorophores belonging to the class of rhodamines or

Experimental results obtained in this section will be applied for the construction of
artificial photonic wire molecules and are subject of the following section of the
discussion. A closer look on energy transfer processes, interchromophoric
interactions and a more mechanistic description will be given.

                                  Results and Discussion

4.2. Design of a Unidirectional Photonic Wire

In the present part of this work, efforts and results towards the realization of synthetic
photonic wires are presented. The objective was to design artificial model
compounds which do transport light energy, similar to electron transport processes in
an electric wire. In contrast to conductive nanowires, these compounds should be
addressed from a distance and in a non-invasive way, i.e. excitation via light.

The first synthetic molecular photonic wires had been realized in 1994 by LINDSEY
and coworkers [Wagner and Lindsey, 1994], based on conjugated porphyrin arrays.
Similar to this approach, photonic wires in the frame of this work were designed with
an optical input unit which serves as light absorbing molecule and transfers its
energy through a number of transfer units, until finally reaching an output unit. This
output unit dispenses energy via light or may be connected to another molecular
system, serving as energy donor.

In contrast to the synthetically complicated porphyrin approach, a different route was
chosen to design molecular photonic wires. Although slightly different in detail, the
principal strategy for the design of such artificial light transporting compounds can be
summarised in the following outline:

          •   DNA was used as rigid scaffold along which fluorophores were
              conjugated using well-developed labelling strategies to obtain well-
              defined geometries and make possible many variations.

          •   Commercially available fluorophores are used as light transmitting
              units, and were chemically conjugated to modified oligonucleotides.

          •   An energy cascade as driving force for the excited-state energy transfer
              was applied to ensure unidirectionality.

          •   The arrangement of fluorophores was realized such that strong
              electronic   interactions   promoting        fluorescence   quenching   are

          •   The distance of fluorophores was chosen to be appropriate for weak
              dipole-dipole induced fluorescence resonance energy transfer (FRET).

                                    Results and Discussion

Because of the unique molecular recognition properties and the scaffold-like
structure, double-stranded (ds) DNA constitutes an ideally suited building block on
which the construction of nanoscale molecular devices can be based [Vyawahare et
al., 2004]. In addition, the use of DNA offers many well-developed labelling and post-
labelling strategies to introduce a variety of different fluorophores in a modular
conception, i.e. short oligonucleotides carrying the desired fluorophores can be
selectively hybridised to a complementary template strand (figure 4-14). The resulting
wire is addressed by excitation of the primary donor, which transfers excited-state
energy according to FÖRSTER theory by weak dipole-dipole induced chromophore
interactions through the transmitter chromophores to the final acceptor. The acceptor
releases the transferred energy by emission of a fluorescence photon.

In the following section, two slightly different approaches based on fluorophores
conjugated covalently to oligonucleotides and exploiting hybridisation of single-
stranded DNA will be discussed. In all model systems, the interchromophoric
distance was chosen to be 10 base pairs, which equals a distance of 3.4 nm. This
ensures efficient energy transfer by the FRET mechanism and prevents electronic
interactions between adjacent chromophores (see (B) in figure 4-14).

     Figure 4-14: Underlying principle for artificial photonic wires. (A) DNA base pairing,
     (B) single strand hybridization forming a photonic wire molecule.

                                   Results and Discussion


The most widely used approach to realize unidirectional photonic wires based on
multi-step fluorescence resonance energy transfer was realized with four single-
stranded DNA fragments of various lengths (60 bases and 20 bases). The DNA
sequence and the building principle is depicted in figure 4-15.

Amino-modified single-strand oligonucleotides were conjugated to a number of
fluorophores according to the protocol presented in the experimental section (see
3.3.1). Purification by gel filtration and HPLC ensured samples only containing the
dye labelled oligonucleotides. Concentrations of samples were measured by steady-
state absorption spectroscopy. Both absorption of DNA (at 260 nm) and fluorophore
were taken into account for determination of concentration. Appropriate fractions of
all four single stranded oligonucleotides were mixed in PBS and hybridised, yielding
60 bp double-stranded DNA with up to five fluorophores at well-defined positions and

To ensure highly efficient fluorescence resonance energy transfer without quenching
electronic interactions, an inter-chromophoric distance of 10 bp corresponding to 3.4
nm was used [Dietrich et al., 2002]. The overall spatial range covered by those
photonic wires reached 13.6 nm at its maximum. The spectral range comprised more
than 270 nm of the visible spectrum, starting from 488 nm to 764 nm in the case of a

    Figure 4-15: Nucleotide sequence of four single DNA strands used in the 60 bp
    approach for synthetic photonic wires, containing up to five fluorophores. Exemplary
    configuration is shown below for Wire03-01.

                                   Results and Discussion

far-red emitting final acceptor. As calculated from the spectroscopic characteristics of
singly labelled oligonucleotides, each FRET step should occur with more than 95%
efficiency, assuming free rotation of the chromophores (κ2 = 2/3). For the exemplary
compound Wire03-01 (see figure 4-13), values obtained for single energy transfer
efficiencies are presented in table 4-9.

                                   ε/l mol-1 cm-1
              λabs/nm   λem/nm                          Φf          R0/Å        EFRET
                                      (x 10 )

   RhG          508       534          0.74             0.9
                                                                     65.0       0.98
   TMR          560       582          0.95             0.9
                                                                     71.7       0.99
 ATTO590        603       625          1.20             0.8
                                                                     72.3       0.99
 ATTO620        622       638          1.20             0.5
                                                                     74.5       0.99
 ATTO680        689       703          1.25             0.3

Table 4-9: Spectroscopic properties, FÖRSTER radii and energy transfer efficiencies for
fluorophores bound to DNA and contributing to the photonic wire Wire03-01.

Assuming an inter-chromophoric distance of 3.4 nm and neglecting the length of
chemical linkers to fluorophores, an overall transfer efficiency of 95% is calculated for
a perfectly aligned wire with the configuration of Rhodamine Green, TMR, ATTO590,
ATTO620 and ATTO680.

Table 4-10 gives an overview of photonic wires which have been synthesized using a
60 bases double-stranded DNA as rigid scaffold. Fluorophores attached to
corresponding single strands are listed, together with the number of energy transfer
steps, NET, and the overall energy transfer efficiencies derived from steady-state
ensemble spectroscopy.

                                       Results and Discussion

     Name                              Fluorophores                                  NET     Efficiency(1)

Wire03-01              RhG, TMR, ATTO590, ATTO620, ATTO680                            4       0.15-0.19

Wire03-02              RhG, TMR, ATTO590, LCR, ATTO680                                4       0.15-0.21

Wire03-03              RhG, TMR, ATTO590, LCR, Cy5.5                                  4       0.13-0.18

Wire04-01(2)           RhG                                                            0            -

Wire04-02              RhG, TMR                                                       1       0.37-0.59

Wire04-03              RhG, TMR, ATTO590                                              2          0.35

Wire04-04              RhG, TMR, ATTO590, LCR                                         3          0.17

Wire04-05              RhG, TMR, ATTO590, LCR, ATTO680                                4          0.12

Wire04-06(3)           RhG, TMR, ATTO590, LCR, ATTO680                                4          0.14

Wire04-07              RhG, TMR, Cy3.5, LCR, ATTO680                                  4          0.05

Wire04-08              RhG, TMR, ATTO590, LCR, Cy5.5                                  4          0.14

Wire04-09              RhG, TMR, ATTO590, ATTO647                                     3          0.37

Wire04-09B(4)          RhG, TMR, ATTO590, ATTO647                                     3          0.34

Wire04-09C             RhG, TMR, ATTO590, ATTO647, ATTO680                            4          0.16

Wire04-10              RhG, TMR, ATTO590, ATTO647, Cy5.5                              4          0.11
Wire04-11              RhG, TMR, ATTO590, ATTO647, ATTO725                            4
Wire04-12              RhG, TMR, ATTO590, ATTO647, ATTO740                            4

Table 4-10: Summary of photonic wire constructs synthesized using the 60bp DNA approach,
NET represents the number of energy transfer steps. RhG = Rhodamine Green, TMR =
Tetramethylrhodamine, LCR = LightCycler Red.                        Energy transfer efficiency from input to
output unit was determined from steady state emission or excitation spectra; if more than one
method was used, a range of values is given. (2)Photonic wires numbered 04-xx exhibit a biotin
linker at the 3’ end of the sense strand,               equals Wire04-06 with different concentration of
sense strand,         slightly different concentrations compared to Wire04-09, for hybridisation
studies with varying temperature only,             these wires were only constructed at the single-
molecule level, no ensemble spectra and energy transfer efficiency are available.

                                  Results and Discussion

Energy transfer efficiencies listed in table 4-10 differ substantially from theoretically
expected values. Photonic wires containing five fluorophores exhibit energy transfer
efficiencies between 13 and 21%, the most efficiently working compound is Wire03-
02 reaching 15 to 21%. It can be anticipated that because of the huge size and
complexity of the system, unfavourable conformational orientations of the
chromophores, insufficient hybridisation, and/or additional quenching pathways
including electron-transfer reactions with DNA nucleotides [Heinlein et al., 2003] will
generate an inhomogeneous broadening of the transfer efficiencies. Nevertheless,
ensemble spectroscopy can give a first estimation of the suitability of a construct.
This is demonstrated by a comparison of energy transfer efficiencies in four dye
photonic wires Wire04-04 (17%) and Wire04-09 (37%) (table 4-10).

Before going more into mechanistic details of artificial photonic wires by applying
single-molecule spectroscopy, different methods suitable to derive energy transfer
efficiencies in multi-chromophoric compounds are the subject of the next paragraph,
including steady-state and time-resolved ensemble spectroscopic methods. The
following considerations only include energy transfer steps between adjacent
fluorophores, the interactions between fluorophores spaced by more than 10 bp is


The quality of energy transfer through an artificial photonic wire is characterized by
its overall energy transfer efficiency. Although this is theoretically easy to determine
from ensemble spectroscopic data for a homogeneous sample exhibiting FRET, a
number of complications occur in photonic wires designed in the described way:

          •   Hybridisation, although being a very selective and efficient process, can
              be difficult if more fragments are involved due to unfavourable
              secondary structure, e.g. kinks, or dye-nucleobase interactions
              [Waggoner and Randolph, 1997].

                                      Results and Discussion

              •   Exact concentrations of single stranded oligonucleotides labelled with
                  fluorophores are difficult to obtain, since extinction coefficients from
                  fluorophores may vary substantially, a result of a altered electronic
                  environment [Haugland, 1989; Clegg et al., 1992b; Han, 2005].

As a result, energy transfer efficiencies from ensemble spectroscopic data essentially
are approximated values, and strongly depend on the heterogeneity of the sample
and the number of energy transfer steps.

Methods to obtain reliable data for energy transfer efficiencies are well-elaborated for
simple FRET-pairs, and include decrease of donor lifetime or quantum yield,
normalized fluorescence emission spectra or anisotropy [Clegg 1992]. Unfortunately,
these methods become extremely uncertain if more than two fluorophores are
participating or larger deviations in concentrations appear. Most work on ensemble
multistep FRET therefore exploited variations in donor emission to determine FRET
efficiencies [Ohya et al., 2003]. Other works leave out the calculation of energy
transfer efficiencies and focus on changes in fractional intensities to derive proximity
data [Clamme and Deniz, 2004]. Especially for three-colour FRET experiments, a
number of sophisticated models were worked out, but model compounds usually are
constructed from one or two components [Liu and Lu, 2002; Watrob et al., 2003]. The
great advantage of single-molecule multi-FRET experiments lies in the observation of
only one multichromophoric system, simplifying data analysis enormously [Hohng et
al., 2004].

To calculate energy transfer efficiencies for photonic wire molecules designed in this
work, a combination of several techniques as used. Especially complex compounds
with more than three fluorophores did not allow the use of time-resolved data, and a
combination of steady-state spectroscopic methods revealed a rough approximation
for energy transfer efficiencies.

              •   Absorption spectra can be expressed as linear combination of
                  contributing single fluorophore spectra. This allows the estimation of the
                  overall stochiometry of the samples. Possible pipetting errors can be
                  revealed yielding unhybridized DNA. Uncertainty remains due to
                  variable extinction coefficients.

                                      Results and Discussion

          •   A linear combination of emission spectra, normalized to the quantum
              yield of each fluorophore, gives a direct estimation of energy transfer
              efficiencies for each step (very similar to the method of sensitised
              emission [Clegg, 1992]).

          •   A comparison of fluorescence photons obtained via energy transfer
              steps and direct excitation of the final emitting fluorophore are a
              measure for the overall energy transfer efficiency. (This method is
              closely related to comparing a measured transfer efficiency to 100%
              transfer efficiency, [Clegg 1992]).

          •   Excitation spectra, recorded at the main emission wavelength from the
              last emitting fluorophore, can be approximated as linear combination of
              the absorption spectra of contributing fluorophores. A normalization to
              the extinction coefficients of the fluorophores yields transfer efficiencies
              for all single steps.

By combining the results obtained in each analysing step, steady-state spectroscopy
yields a sufficient approximation of energy transfer efficiencies. An exemplary
calculation, using spectroscopic data shown in figure 4-16, will be presented for the
system Wire03-01 in the following. These methods were used for the determination
of energy transfer efficiencies listed in table 4-10.

                                   Results and Discussion

Figure 4-16: Overview of ensemble-spectroscopic steady-state methods to characterize
photonic wire molecules, exemplary for the compound Wire03-01 with 5 fluorophores. (A)
Absorption spectrum exhibiting four peaks (two dyes are contributing to the 618 nm
absorption peak); (B) emission (488 nm excitation) and excitation spectrum (fluorescence
recorded at 720 nm); (C) and (D): the measured overall absorption spectrum and emission
spectrum is calculated as linear combination of single fluorophore spectra; (E) emission of
the last acceptor excited via energy transfer and directly; (F) excitation spectrum calculated
from single fluorophore absorption spectra.

                                      Results and Discussion

Steady-State Absorption Spectra

In a first step, the absorption spectrum of a photonic wire sample, A(λ), is
approximated by single fluorophore absorption spectra, ai(λ),

                                 A(λ ) = ∑ α i , rel ai (λ )                             ( 4-2 )

Since both absorption wavelength and extinction coefficient slightly change, these
values are necessarily approximated. Values obtained for the approximation of the
absorption spectrum of Wire03-01 are listed in table 4-11.

                            RhG             TMR           ATTO590 ATTO620 ATTO680

      αrel                  0.27            0.35                0.47     0.47     0.50

      εmax (l mol-1 cm-1)   74000          95000               120000   120000   125000

      αabs                  0.91            0.92                0.98     0.97     1.00

Table 4-11: Linear combination of single fluorophore absorption spectra to approximate the
absorption spectrum of Wire03-01, normalized to the extinction coefficient.

In table 4-11, αrel represents the relative contribution to the overall absorption of the
hybridised photonic wire, which was normalized to a maximum value of 1 (see part
(C) in figure 4-16). Corrected by the extinction coefficient of each fluorophore, εmax,
an absolute contribution or ratio is obtained, αabs, which was normalized with respect
to the maximum ratio appearing.

Regarding the data presented in table 4-11, a few issues are worth mentioning. First,
ratios of fluorophores are not equal. The reason is a combination of uncertainties in
concentration determination, insufficient purification steps, and errors made when
pipetting single strand volumes for hybridisation. Total differences of ~ 10% in
concentration are observed. Secondly, both TMR and ATTO590 are conjugated to
the same strand, so their relative contribution is expected to be identical. In this case,
a slight change in the extinction coefficient can be assumed.

                                        Results and Discussion

Steady-State Emission Spectra

In a second step, the overall emission spectrum of a photonic wire, F(λ), is
approximated by a linear combination of single emission spectra of contributing
fluorophores bound to DNA, fi(λ),

                                   F (λ ) = ∑ α i , rel f i (λ )                                     ( 4-3 )

This calculation is a variation of the often used method of sensitised emission [Clegg
1992], values obtained for the linear combination of Wire03-01 are listed in table 4-
12, fluorescence emission spectra are shown in figure 4-16, (D).

                       RhG               TMR               ATTO590        ATTO620         ATTO680

   αrel                1.00               0.20                 0.27           0.60            0.15

   φ                    0.9                0.9                     0.8        0.5             0.3

   αabs                1.11               0.22                 0.34           1.20            0.50

   αabs, norm          0.33               0.06                 0.10           0.36            0.15

Table 4-12: Linear combination to approximate the emission spectrum of Wire03-01 by relative
contributions of single fluorophores, αrel. Normalized to the quantum yield, φ, absolute
contributions αabs are obtained and normalized to the sum of 1, αabs,norm, now giving a measure
of energy transfer at each step.

Relative contributions to the overall emission spectrum of the sample Wire03-01
must be normalized to the quantum yield of each individual fluorophore to obtain
absolute contributions, αabs. This representation now allows to estimate energy
transfer   efficiencies    by      analysing        the      “distribution”   of     energy   along     the
multichromophoric sample. The value αabs for the last emitter, ATTO680, gives a
value of 15% for the total energy transfer observed.

The method of approximating multi-step FRET emission spectra is only reliable if the
quantum yield of fluorophores in the hybridised compound is comparable to the
quantum yield of fluorophores attached to single strand DNA. Excitation of the
second fluorophore or skipping of transfer steps are not taken into account, since a
possible better accuracy would not compensate greater sources of errors.

                                       Results and Discussion

Furthermore, the absorption spectrum should exhibit equal contributions of

Remarkably, this approximation already reveals a major problem of photonic wire
molecules designed from several single stranded DNA molecules exploiting
hybridization, showing energy loss at two important positions: the first dye and input
unit, and the fourth dye in the cascade.

Direct Excitation of the Last Emitter

Another method applied to evaluate the overall transfer efficiency is illustrated in part
(E) of figure 4-16. The approach is related to the method of comparing the acceptor
emission of a sample to a second sample exhibiting 100% FRET [Clegg 1992]. As no
sample with 100% FRET is available with the present design, fluorescence emission
of the last emitting unit was compared to direct excitation of the fluorophore. For
comparability of both values, differences in the excitation probability of a fluorophore
must be taken into account, which is realized by normalizing fluorescence intensity to
the extinction coefficient.

An exemplary calculation for the sample Wire03-01 (see figure 4-16, (E)) is
summarized in table 4-13.

                                      Excitation via FRET (488 nm)      Direct Excitation (685 nm)

Fluorescence emission at 700                        14                                 113
nm (counts)

Contributing     emission     from                 8.1                                 113
ATTO680 (counts)

Extinction      coefficient      at               42000                           119000
excitation wavelength (l mol-1

Normalized emission (counts)                       20.3                                100

Table 4-13: Calculation of energy transfer by comparing fluorescence emission along the
FRET-system and direct excitation of the last emitter. Fluorescence emission is corrected for
preceding fluorophores in Wire03-01 and for the extinction coefficient of the excitation
wavelength; finally, the value for direct excitation is set to 100 for more clarity.

                                    Results and Discussion

Excitation of Wire03-01 at 488 nm yields a fluorescence emission intensity of the last
emitter (measured at 700 nm) of 14 counts via energy transfer and 113 counts via
direct excitation, respectively (685 nm excitation). Fluorescence emission of
ATTO680 derived form FRET-excitation has to be corrected, i.e. the overlap with
preceding fluorophores has to be ruled out. In a final step, extinction coefficients at
the excitation wavelength have to be included, yielding a final value of 21% for
energy transfer in Wire03-01.

This approach is very sensitive to an excess of concentration of the final dye which is
not involved in the formation of double strand DNA, e.g. due to concentration errors
or incomplete hybridisation.

Steady-State Excitation Spectra

The last method makes use of the excitation spectrum of a photonic wire sample,
recorded at the emission wavelength of the last emitting dye in the energy transfer
cascade. This method is not reliable if an excess of the last fluorophore is present.
The excitation spectrum is approximated by a linear combination of single dye
absorption spectra, yielding relative contributions of participating fluorophores which
contribute to the emission of the last fluorophore, see (F) in figure 4-16. Exemplary
calculations for Wire03-01 are listed table in 4-14. Note that a normalization to
quantum yields is not necessary, since only one emission wavelength which
originates mostly from the last emitter is observed.

                       RhG          TMR            ATTO590     ATTO620        ATTO680

   αrel                0.10          0.20           0.55          0.23           0.88

   εmax (l mol-1       74000        95000          120000       120000         125000

   αabs                0.19          0.30           0.65          0.27           1.00

Table 4-14: Energy transfer efficiency derived from linear combination of excitation spectrum.
Relative contributions αrel are normalized to the extinction coefficient, yielding absolute
contributions, αabs.

                                   Results and Discussion

From the relative contributions of fluorophores to the recorded excitation spectra of
Wire03-01, absolute values can be obtained by normalizing to the extinction
coefficients of the dyes. These absolute values αabs are then normalized to the value
of 1 for the final fluorophore, ATTO680. As a result, the excitation efficiency of
ATTO680 along the energy transfer cascade is 19% with respect to the first
fluorophore RhG.

Three different models of calculating energy transfer efficiencies were demonstrated
for the sample Wire03-01 and yielded values between 15% and 21% (and a mean
value of 18%). The combination of steady-state spectroscopic techniques allows to a
certain extent to characterize the quality of a sample as artificial photonic wire, but
implies a number of difficulties. The number of five spectrally close fluorophores
prevents to account for unequal stoichiometry of contributing oligonucleotides, and
subpopulations cannot be identified. At this step, it becomes clear that more
sensitised methods must be applied, which allow to watch single photonic wires at
their working level, i.e. the single-molecule level.

                                  Results and Discussion


Besides ensemble spectroscopic data derived from steady-state measurements,
time-resolved data can be consulted to determine energy transfer efficiencies [Clegg,
1992]. This is particularly feasible in “classic” FRET-pairs, constituted of two
fluorophores, although often less precise due to a complex environment of
fluorophores attached to DNA which may lead to deactivation processes influencing
the fluorescence lifetime [Dietrich et al., 2002]. Nethertheless, subpopulations with
different kinetic parameters can be distinguished and quantified, which allows a more
detailed description beyond the possibilities of steady-state spectra.

As an example for determination of energy transfer efficiency based on time-resolved
measurements, the sample denoted Wire04-02 is discussed in the following. It
consists of only two fluorophores with a distance of 10 base pairs or 3.4 nm,
Rhodamine Green and TMR, respectively. The free dyes exhibit a fluorescence

 Figure 4-17: Time-resolved fluorescence measurements of the FRET-pair RhG-TMR, spaced
 by 10 base pairs of dsDNA (excitation at 495 nm, 2000 counts in maximum channel, time
 resolution 0.012 ns/channel). (A): donor emission (green) compared to free donor (blue)
 recorded at 535 nm, lamp profile (grey); (B) Acceptor emission (red) and free acceptor
 (green) recorded at 585 nm.

                                   Results and Discussion

lifetime of 3.7 ns for RhG and 2.2 ns for TMR, respectively. Measurements were
performed at a concentration of ~10-6 M in PBS using a time-resolved spectrometer
with an LED as excitation source, emitting at 495 nm. Fluorescence emission from
the donor fluorophore was recorded at 535 nm, acceptor fluorescence at 585 nm. For
a better comparability, fluorescence decays of both donor and acceptor only, both
conjugated to the same double-stranded DNA, were measured as well. Fluorescence
decays and the lamp profile are shown in figure 4-17. From a first view, the decay of
the donor fluorophore shows a short component with respect to the free fluorophore,
which results from a fast deactivation of the excited state by energy transfer. This is
predominantly the case for short times of the decay. Going to longer times, both
decays appear parallel and exhibit similar exponential kinetics, which indicates a
second subpopulation in the FRET-pair sample of Wire04-02 that does not exhibit
energy transfer. This can be confirmed by the results obtained after deconvolution
and fit procedures, listed in table 4-15.

                                            τ1 / ns   A1     τ2 / ns   A2      χ2

         Rhodamine Green                     4.20     1.00      -       -     1.154

         Rhodamine Green (donor)             0.70     0.23    3.66     0.77   1.084

Table 4-15: Fluorescence kinetics from Rhodamine Green bound to dsDNA in the presence and
absence of an acceptor molecule.

The results from the fit procedure for Rhodamine Green in the presence of an
acceptor fluorophore confirms the first impression: two subpopulations can be
distinguished. A first fraction shows high efficient FRET with a transfer efficiency of E
= 0.83 and an amplitude of A = 0.23. A second fraction of 77% exhibits a
fluorescence lifetime which is comparable to the free dye, but still a difference of 0.54
ns is observed. If calculating a FRET-efficiency, a value of E = 0.13 is obtained. At
this point, it is difficult to further characterize this second subpopulation. A more
detailed picture of Rhodamine Green as donor fluorophore attached to DNA will be
given from single-molecule results later in this work.

The decays in the acceptor channel are more difficult to interpret. Results from the fit
procedures are listed in table 4-16.

                                 Results and Discussion

                                       τ1 / ns    A1      τ2 / ns   A2      χ2

          TMR                           3.45     0.85      0.69     0.15   1.126

          TMR (acceptor)                2.25     0.48      4.39     0.52   1.092

Table 4-16: Fluorescence kinetics calculated for TMR bound to DNA and as the acceptor in

Without the presence of a donor fluorophore, TMR exhibits two fluorescence lifetimes
of 3.45 ns (85%) and 0.69 ns (15%). This behaviour can be explained by efficient
dynamic quenching of fluorescence by a neighbouring guanosine residue [Eggeling
et al., 1998b]. If a donor fluorophore is present, one would expect a broadened
fluorescence decay of the acceptor around the maximum, which is slightly shifted
with respect to the decay of the acceptor fluorophore without the presence of a
donor. The reason for the appearance of a so-called rise time lies in the “excitation
profile” for the acceptor: unlike the donor whose fluorescence decay is convoluted
with the lamp profile, the decay of the acceptor is additionally convoluted with the
fluorescence decay of the donor. In mathematical analysis by deconvolution and
multi-exponential fitting, this usually yields a short component with a negative
amplitude. The absolute value of this negative amplitude determines the fraction of
acceptor molecules which is excited through energy transfer, and the rise time itself
allows to determine energy transfer efficiencies, complementary to the procedure
applied for the donor fluorophore. Regarding fluorescence decays in figure 4-17, the
rise time can only be anticipated from the shape of the acceptor decay, but cannot be
resolved mathematically. The values obtained suggest many contributions to
fluorescence at 585 nm and do not allow an exact interpretation. Direct excitation of
the acceptor (~20 % at 495 nm for TMR) and remaining fluorescence of the donor
(~18% for Rhodamine Green) generally make the determination of energy transfer
efficiencies very difficult, which explains the more common use of donor fluorescence

It is now interesting to compare the results obtained from time-resolved ensemble
measurements to steady-state measurements performed under equal conditions for
the same sample. Steady-state absorption, emission and excitation spectra are
depicted in figure 4-18.

                                      Results and Discussion

 Figure 4-18: Steady-spectra of the FRET-pair RhG-TMR (Wire04-02): (A) absorption spectra,
 (B) emission spectra, (C) excitation by two wavelengths (488 nm and 560 nm), (D) excitation

First, the absorption spectrum of the compound Wire 04-02 is approximated by linear
combination of single fluorophore spectra. Values obtained are listed in table 4-17.

                                                RhG            TMR

                       αrel                     0.84           0.73

                        εmax (l mol-1 cm-1)    74000           95000

                       αabs                     1.00           0.74

            Table 4-17: Linear combination of the absorption spectrum of Wire04-02.

The linear combination of table 4-17 reveals an excess of 26% for Rhodamine Green
in Wire04-02. Using previously presented approaches to calculate the energy

                                    Results and Discussion

transfer efficiencies, values obtained from different emission spectra are summarized
in table 4-18. Concentration errors were taken into account.


                   Emission spectra                              0.46

                   Direct excitation of acceptor                 0.59

                   Excitation Spectra                            0.37

  Table 4-18: Energy transfer efficiency calculated for Wire04-02 applying different methods.

All three methods used for calculation of energy transfer efficiencies yield varying
results, and a mean value of 47% can be given. To compare this value with the result
from time-resolved spectroscopy, a surplus of 26% of RhG must be included (the
method of time-resolved spectroscopy is more sensitive to a higher donor
concentration than the method of excitation spectra or direct excitation of acceptor).
By this, the energy transfer efficiency is reduced to 35%, whereas time-resolved
measurements yield 19% (if only the first population with very efficient FRET is taken
into account) or 29% (with both distributions) as a mean value, respectively. As a
result, it can be stated that absolute values for transfer efficiencies are quite different.
This ambiguity between time-resolved and steady-state spectroscopy has already
been observed previously [Dietrich et al., 2002].

Time-Resolved Emission Spectroscopy of Photonic Wires

As an extension to mere time-resolved measurements of fluorescence kinetics at
fixed wavelengths, an impression of the complexity of energy transfer along a
photonic wire with five fluorophores can be obtained from spectrally resolved
fluorescence decays. This was realized by excitation at 495 nm and recording
fluorescence decays in steps of 10 nm, yielding a time-resolved emission spectrum
(TRES). In part (A) of figure 4-19, a two-dimensional projection of a TRES spectrum
from Wire03-01 ranging from 520 nm to 740 nm detection wavelengths is depicted,
and a number of spectroscopic properties can be read out immediately from such
presentations. First, short fluorescence lifetimes are observed in the donor region
(around 550 nm), whereas in the acceptor region (> 630 nm), fluorescence lifetimes
                                    Results and Discussion

 Figure 4-19: (A) Time-resolved emission spectrum of Wire03-01, recorded in 10 nm steps
 from 520 nm to 740 nm (2000 maximum counts, 495 nm excitation, channel resolution 0.012
 ns). (B) Magnification shows a shift of the area of maximum photon counts towards longer
 time channels of 0.84 ns at its maximum. (C) Fluorescence decays of the four most
 important emission wavelengths of Wire03-01, (D) calculated fluorescence kinetics after
 deconvolution and biexponential fitting.

are considerably longer. A second issue which is emphasized by black circles in part
(B) of figure 4-19 is the broadening and shifting of the fluorescence decay around its
maximum. With a channel length of 12 ps, a shift of 70 channels or 0.84 ns is
observed. Both shifting and broadening are due to rise times that appear in each step
of energy transfer. In part (C), four selected decays together with the lamp signal
show the behaviour of both broadening and shifting for the most important emission
wavelengths of the sample, Wire03-01.

Deconvolution and fit procedures (see part (D), figure 4-19) yield strongly quenched
subpopulations at shorter wavelengths, whereas at longer emission wavelengths, a
rise time with a contribution of 24% appears.

                                  Results and Discussion

Interestingly, shortest times and largest amplitudes are not observed around the
emission wavelength of the first fluorophore, but at ~590 nm. This may be explained
by a fraction of donor fluorophores which fluoresces without any energy transfer
process, comparably to previously presented results for the FRET-pair RhG-TMR. On
the other hand FRET from TMR to ATTO590 is very efficient and quantitative.

In Figure 4-20, fluorescence emission spectra and time-resolved spectra of two
compounds denoted Wire04-09 and Wire04-10 are depicted. Both are build up from
rhodamine dyes (RhG, TMR, ATTO590) and a carborhodamine dye on position 4
(ATTO647). Additionally, Wire04-10 has a dye at position 5, the carbocyanine dye
Cy5.5 which exhibits a fluorescence lifetime of ~1.0 ns. Spectra are recorded from
530 nm to 690 nm and 720 nm, respectively. In the case of Wire04-09, a broader
“shoulder” appears around 640 nm. This “shoulder” is depopulated by energy transfer
towards Cy5.5 in Wire04-10. Due to a significantly shorter lifetime, the broadening
becomes smaller and demonstrates the presence of a subpopulation which exhibits
energy transfer to the last fluorophore.

 Figure 4-20: Ensemble and time-resolved emission spectra of photonic wires Wire04-09 (530
 to 690 nm) and Wire04-10 (530 to 720 nm). 495 nm excitation, 2000 maximum counts, 0.012
 ns / channel.

                                 Results and Discussion

Compared to Wire03-01 (figure 4-19), a shift of fluorescence decays of 1.07 ns is
observed for Wire04-09 (figure 4-20). This underlines a higher efficiency in energy
transfer which was already calculated from steady-state ensemble spectra (energy
transfer efficiency is determined to be 37%, see table 4-10). In Wire04-10, a shift of
0.70 ns is observed at 720 nm (mainly fluorescence emission from Cy5.5), together
with 0.97 ns at 670 nm (maximum emission wavelength of the fluorophore
ATTO647). This observation is in good agreement with steady-state spectroscopy,
where Wire04-10 only yields 11% of energy transfer efficiency (in other words, only
~30% of energy transfer from ATTO647 to Cy5.5 is observed). This observation is in
agreement with the emission spectrum of Wire04-10 (figure 4-20), which exhibits a
weak contribution from the last chromophore (Cy5.5), but still considerable
contribution from the fourth dye (ATTO647).

Both Wire04-09 and Wire04-10 will further be characterized in single-molecule
experiments in aqueous environment.

4.3. Studying Photonic Wires with Single-Molecule Spectroscopy

It is the complexity of the systems - independent of the energy-transfer mechanism
employed - which determines the need for new analytical techniques for the
characterisation of bottom-up nanotechnological devices, such as photonic wires.
Single-molecule fluorescence spectroscopy (SMFS) is a technique that provides
detailed information required for the analysis of static heterogeneity. In addition,
SMFS also enables to probe the quality of the device.

Molecular photonic wires have to operate at the single-molecule level and, hence,
they have to be characterised at this individual level as well. This was realized by
using two principal techniques of immobilization. At first, fluorescence imaging and
traces of individual photonic wires were derived from molecules adsorbed on a dried
glass substrate. The second technique made use of biotin-streptavidin interaction
and allowed immobilization in liquid environment, which is of particular interest if the
chemical properties of the surroundings are object of change.

                                      Results and Discussion


Spectral information of photons emitted from photonic wire samples is detected with
four APDs as detector channels. A set of dichroic beamsplitters and filters hereby
selects a spectral region. At the single-molecule level, it is possible to assign a
photon to a certain emitter, by exploring its spectral characteristics.

To get an idea of the origin of a photon, i.e. to assign probabilities which dye
contributed a particular detected photon, two approaches to obtain the spectral
distribution of a fluorophore onto the four detectors were used. At first, solutions of all
fluorophores bound to DNA were excited with 488 nm and fluorescence was
collected independently. To excite longer wavelength absorbing dyes with the
excitation source, higher concentrations were used to compensate a lower
absorption probability (it is assumed that the fluorescence emission of these
fluorophores    does    not   alter    with   the    excitation   wavelength).   In     general,
concentrations between 10-8 M and 10-6 M were used. This minimizes contributions
from Raman scattering around 570 nm, originating from symmetric and asymmetric
vibrations of water molecules. Fluorescence emission of the five chromophores from
Wire03-02 broken down into detector channels are presented in table 4-19. Detector
channels are denoted according to their approximated spectral detection area, i.e.
blue, green, yellow and red.

                       Blue               Green                Yellow            Red

   RhG                 0.58                0.37                 0.05             0.01

   TMR                 0.02                0.72                 0.22             0.03

   ATTO590             0.01                0.33                 0.54             0.12

   LCR                 0.06                0.14                 0.61             0.19

   ATTO680             0.02                0.08                 0.08             0.82

Table 4-19: Measured relative spectral distribution from five fluorophores constituting Wire03-
02 onto four detector channels.

                                   Results and Discussion

In a second approach, the fluorescence emission spectra of each fluorophore was
multiplied with the transmission curves of the detectors (see section 3.1 and figure 3-
5, 3-6). Spectrally separated detection patterns for the fluorophores discussed in this
section are depicted in figure 4-21.

 Figure 4-21: Calculated distributions of fluorescence spectra on the four detector channels
 from five fluorophores constituting the sample Wire03-02, i.e. Rhodamine Green, TMR,
 ATTO590, LCR and ATTO680.

After integration and normalization of the calculated patterns, theoretical spectral
distributions are derived. Values are listed in table 4-20.

                     Blue                Green              Yellow             Red

  RhG                 0.68               0.28                0.03              0.00

  TMR                 0.02               0.75                0.21              0.02

  ATTO590             0.00               0.07                0.82              0.11

  LCR                 0.00               0.01                0.86              0.14

  ATTO680             0.00               0.00                0.05              0.95

Table 4-20: Relative spectral distribution (calculated) from five fluorophores constituting
Wire03-02 onto four detector channels.

                                  Results and Discussion

Differences between the measured and calculated spectral distributions do appear, if
tables 4-19 and 4-20 are compared. This should be viewed from the point that
measured values are obtained using a configured set-up, whereas calculated values
were obtained from individual absorption and emission spectra of set-up
components, e.g. filters and dichroic beamsplitters, together with the quantum
efficiency of an APD. Sources of error that are to be mentioned are the polarization
dependence of transmission properties from dichroic beamsplitters, small deviations
from a perfect geometrical arrangement of all four detectors or possible optical
aberrations along the detection pathway.

In the following sections, all spectral considerations used in the interpretation of
single-molecule data are based on the measured distributions of fluorescence.


First single-molecule measurements of photonic wire compounds were realized at a
glass-air surface [Heilemann et al., 2004]. Sample molecules were randomly
adsorbed on a dried glass surface and scanned. To provide intensity scan images
with more information, the spectral characteristics of emitted photons is displayed in
scan images as false colour code. This is realized by assigning a colour bit, i.e. blue,
green, yellow or red, to a certain x,y-position and calculate a colour value by
weighting the intensity of each channel. As a result, a false colour byte is obtained
and allows for a first impression from which unit of
a photonic wire a photon originates. Since four
detector channels are used, the representation as
false colour image is ambiguous. In its classical
use, false colour information is created using red,
green and blue (RGB) colour bits, resulting in the
combination of colours yellow, magenta and cyan
(see figure 4-22). If yellow is used as fourth colour
                                                           Figure 4-22: RGB scheme to
and independently, additional combination colours
                                                           produce false colours. A mixture
do appear. Nevertheless, this method is applied for        of red, green and blue yields
a first visualization of the spectral distribution of      white.
fluorescence data, and exact evaluation requires a

                                   Results and Discussion

closer look.

An exemplary scan image is shown in figure 4-23. The magnification of two individual
spots gives a nice impression of the usefulness of false colour images in
multichromophoric systems, revealing photophysical reactions of photonic wire
molecules which occur on a millisecond timescale.

 Figure 4-23: 10x10 µm scan image of Wire03-02, represented as false colour image (2 ms
 integration time, 10 – 100 counts / 2 ms). The sample is composed of the fluorophores
 Rhodamine Green, TMR, ATTO590, LightCycler Red, ATTO680. Two magnified spots show
 spectral dynamics. The histogram on the right side compares the theoretical spectral
 distribution of the sample derived from ensemble spectroscopy (grey columns) to single
 molecule data, derived from about 200 photonic wire spots.

As indicated in figure 4-23, the majority of fluorescence spots is dominated by the
emission of one of the five chromophores. Besides unfavourable conformations and
competing quenching pathways, premature photobleaching of chromophores can as
well substantially control the observed EET pathways and efficiencies. On the other
hand, about 10% of all photonic wires show predominately emission on the red
channel, that is, highly efficient (up to 90%) multistep EET across 13.6 nm. Single
spots are often heterogeneous due to photobleaching or due to reversible
photophysical and chemical reactions.

To better characterise spectral properties of fluorescence emission and compare
single-molecule results to ensemble measurements, ~200 fluorescence spots were
chosen and histogrammed according to their individual spectral pattern (see figure 4-
23, coloured bars in histogram). In a second step, an ensemble fluorescence
emission spectrum of a 10-7 M solution of the photonic wire was taken to derive the
                                    Results and Discussion

spectral ensemble pattern on four detectors (grey bars in histogram). With the
exception of the blue channel, the histogram corresponds well to the ensemble
fluorescence spectrum of the photonic wire.

The deviation found for the blue channel is a result of the lower photostability and a
second red-shifted emissive state of the first donor RhG when immobilized on dry
glass surface. About 60-70% of the RhG molecules show this red-shifted state before
complete photodestruction. These observations for RhG are summarized in figure 4-
24. Part (A) shows the normalized absorption and fluorescence emission spectrum of

 Figure   4-24:   (A)   Ensemble   absorption    and   emission   spectra   of   RhG   labelled
 oligonucleotides, (B) False colour fluorescence intensity image (20 x 20 µm ) of RhG
 labelled oligonucleotides adsorbed on dry glass surface (488 nm excitation, 50 nm/pixel, 2
 ms integration time, 5-60 counts/2 ms). (C) Fluorescence intensity trajectory of an
 individual RhG labelled oligonucleotide measured on the four spectrally separated APDs (3
 kW/cm2 average excitation intensity at 488 nm). (D) Spectral distribution histogram of
 individual RhG fluorescence spots (coloured bars) compared to ensemble pattern (grey

                                 Results and Discussion

a 10-6 M solution of RhG in PBS, exhibiting a maximum at 509 nm and 534 nm,
respectively. However, if absorbed onto a glass surface, an important fraction shows
red-shifted emission, demonstrated in the false colour scan image (15 x 15 µm²) in
figure 4-24, (B). An exemplary fluorescence trajectory affirms this observation ((C),
figure 4-24): while the spectral pattern of emission during the first ~1.5 s matches the
ensemble emission spectrum, a spectral shift towards longer emission wavelengths
is observed before photobleaching at ~2.5 s. To quantify this observation and
compare the spectral behaviour of RhG in ensemble and single-molecule
measurements performed on dry glass, a similar spectral pattern as in figure 4-23
was elaborated, shown in figure 4-24 (D). The histogram reflects the tendency of red-
shifted fluorescence observed for single RhG molecules adsorbed randomly on dried
glass in a nice way. Interestingly, this behaviour was not observed for single RhG
molecules immobilized under aqueous conditions (see 4.3.3).

Better insight into dynamic behaviour can be obtained from fluorescence trajectories
of individual photonic wires. The subpopulation of wires that initially emits
predominately on the red channel exhibits up to five successive photobleaching
events. Sequential photobleaching accompanied by a shift in the emission spectrum
from the red back toward the blue confirms that the energy is transferred stepwise
among the chromophores, that is, subsequent unidirectional four-step EET
[Heilemann et al., 2004]. An exemplary fluorescence trajectory exhibiting all five
steps is depicted in figure 4-25. The spectral information provided by fluorescence
photons is integrated for each emission step (see histograms in figure 4-25, coloured
bars) and compared to the single fluorophore’s spectral signature derived previously
(see section 4.3.1; grey bars in histograms).

According to these patterns, five regions with different spectral characteristics can be
retrieved from the four different fluorescence intensity trajectories (figure 4-25). The
first 0.4 s of the trajectory (part 1) are dominated by the emission of the far-red
chromophore ATTO680. Here, the transfer efficiency reaches a value of ~90%. After
photobleaching of the final acceptor, the fluorescence emission is dominated by LCR
(part 2) followed by ATTO590 (part 3), TMR (part 4), and finally Rhodamine Green
(part 5). Overall, five subsequent photobleaching events can be uncovered.
Comparison of the intensity patterns measured (colour columns) to the ones
expected for the different chromophores (grey columns) demonstrates that the
fluorescence emission is dominated by one chromophore at all times. Small
                                   Results and Discussion

 Figure 4-25: Fluorescence intensity trajectory of an individual photonic wire measured on
 the four spectrally separated APDs (3 kW/cm2 average excitation intensity at 488 nm). Five
 different emission patterns can be observed indicating that subsequent photobleaching of
 the chromophores starting with ATTO680 (part 1) occurred. In the lower part, intensity
 patterns (colour columns) measured for the five different parts are compared to those
 measured for singly labelled oligonucleotides (grey columns).

deviations, e.g., those observed for part 1 in figure 4-25, indicate less efficient energy
transfer between some chromophores, i.e., a leakage of the wire, possibly induced
by unfavourable orientations of some fluorophores.

Many trajectories, however, exhibit a more complex photophysical behaviour and do
not show five successive photobleaching events. The trajectory of the photonic wire

                                  Results and Discussion

shown in figure 4-26 (A), for example, initially exhibits efficient EET to the final
acceptor ATTO680 with an efficiency of ~70 %. Further on, the trajectory is
characterised by collective off-states of the whole wire and fluctuations in EET
efficiency along the wire. Collective off-states have frequently been observed in multi-
chromophoric systems and are ascribed to quenching of the fluorescence by
nonfluorescent traps such as, for example, triplet states or radical ions.[Yu et al.,
2000; Tinnefeld et al., 2003; Hofkens et al., 2000; Vosch et al., 2003].

It is interesting to note that the collective off-states, which are most likely caused by
traps located on the final acceptor ATTO680, exhibit a higher quenching efficiency for
the other fluorophores than for the active ATTO680. In addition, the fluorescence of
the wire does not cease by successive bleaching events but some bleaching steps
are lacking (e.g., the part dominated by TMR emission at about 5.8 s is missing,
figure 4-26 (A)).

 Figure 4-26: Fluorescence trajectories (fluorescence photons, νF, in 10 ms versus time in
 seconds) of two photonic wire. Photophysical dynamics beyond bleaching of the
 chromophores are evident.

As another example, the fluorescence trajectory (B) in figure 4-26 starts with
emission from ATTO680, but misses the next step and directly shows a spectral
pattern which resembles a mixed signature from ATTO590 and TMR. After a long
“off”-period, fluorescence from TMR is recovered and finally followed by Rhodamine

                                   Results and Discussion

The complex behaviour of the photonic wires is not necessarily surprising as the
photophysics and photochemistry of the wire do not have to represent the sum of the
behaviour of the individual chromophores; for example, EET in the weak coulombic
regime is not restricted to the nonradiative transfer of energy from a donor in the
excited state to a ground-state acceptor. Transfer processes that are allowed within
the FÖRSTER formalism are those for which there are no changes in electron spin in
the acceptor transition. Hence, the transfer of excitation energy from a chromophore
residing in the first excited singlet state to another chromophore residing either in the
triplet or in an excited singlet state are possible and competitive energy-transfer
pathways, which have far-reaching impact on the photophysics and photochemistry
of multichromophoric systems, such as enhanced intersystem crossing and reduced
photostability [Hofkens et al., 2003; Tinnefeld et al., 2004].


An improved approach to try to control the enormous inhomogeneous broadening
observed in single-molecule measurements of artificial photonic wire molecules was
realized by immobilization photonic wires onto a coated glass surface under aqueous
conditions (see 3.3.2). Samples that are denoted Wire04-xx were equipped with a
biotin linker and allowed binding onto a streptavidin coated surface under aqueous
conditions. Photonic wires were predominantly constructed using rhodamines
(Rhodamine Green, TMR, ATTO590) and carborhodamines (ATTO647) in reductive
environment by adding around 100 µl of 1 M solution of MEA. As discussed
previously, this treatment reduces “off”-times and enhances photostability of this
class of fluorophores and allows longer observation times. Since long-wavelength
fluorophores with rhodamine-like structure are not available for conjugation
chemistry, final acceptor dyes were carbocyanines (Cy5.5), oxazines (ATTO680) or
carbopyronines (ATTO725, ATTO740). Scan images from six different photonic wires
are portrayed in figure 4-27.

The left image in the upper row of figure 4-27 shows a 8 x 8 µm2 false colour scan
from Wire04-01, which only contains one fluorophore, Rhodamine Green. The whole
image demonstrate a higher homogeneity than observed on glass (see 4.3.2), which
is explained by a very similar environment for each fluorophore. It can be noted that

                                     Results and Discussion

     Figure 4-27: 8 µm x 8 µm false colour scan images of photonic wire compounds
     immobilized in PBS buffer with MEA added (excitation wavelength 488 nm,
     integration time 2 ms/pixel, intensity 20 – 200 counts/2 ms). Upper row: Wire 04-01,
     Wire04-02, Wire04-03, lower row: Wire04-09, Wire04-10, Wire04-11.

no spectral shifts of fluorescence are observed, as previously discussed when the
fluorophore is adsorbed on dry glass. The second image and the third image in figure
4-27 represent false colour scan images of Wire04-02 and Wire04-03, respectively.
The false colour information allows in both images to distinguish between the final
emitting fluorophore, which is either RhG, TMR or ATTO590. In the lower row, scan
images of the samples Wire04-09, Wire04-10 and Wire04-11 are shown.

Due to the higher homogeneity observed in aqueous solution, fluorescence emission
of single photonic wires is analysed by fractional intensity distributions. The spectral
information which is encoded in the false colour images can be extracted by
calculating the fractional intensity for each pixel which is above a certain threshold
and belongs to a PSF of a molecule. The fractional intensity for different spectral
signatures is hereby expressed as

                                                      I green
                             F2( green,blue ) =                                        ( 4-4 )
                                                  I green + I blue

                                       Results and Discussion

                                                            I yellow
                            F2( yellow, green ) =                                                      ( 4-5 )
                                                    I yellow + I green

                                                            I red
                             F2( red , yellow) =                                                       ( 4-6 )
                                                    I red   + I yellow

If we assume that a fluorophore predominantly emits on two detector channels, each
fluorophore gets a signature by at least one of these values which allows to
discriminate between different emitters.

Histograms for fractional intensities obtained for Wire04-01 , Wire 04-02 and Wire04-
03 are depicted in figure 4-28 and were calculated from 100-300 individual
fluorescence spots each. To derive characteristic values for each fluorophore, the
mean value of the distribution was determined. Any symmetric fit function, e.g.
Gaussian or Lorentzian functions, is not suitable, since the definition of fractional
intensities is naturally asymmetric. For Wire04-01 (image (A) in figure 4-28), a small
spectral pattern for Rhodamine Green as single emitter in this construct is observed,
characterized by a mean value of F2 (green, blue) = 0.30. In the case of Wire04-02, which
is constituted of two fluorophores, Rhodamine Green and TMR, the same signature
is observed for the donor dye, and the acceptor shows a well separate distribution
with a mean value of F2 (green, blue) = 0.67 (image (B) in figure 4-28). Additionally, the
fractional intensity of the yellow and green channel yields a value of F2 (yellow, green) =
0.42 (image C). Adding one more fluorophore for Wire04-03, the fractional intensity
exhibits the previously determined value for Rhodamine Green on the green and blue
channel (image D), but shows a broader distribution for F2                       (yellow, green),   which is a
combination of ATTO590 and TMR (image E). The last emitter of this compound can
be identified in the histogram of F2         (red, yellow)       with a mean value of 0.21. Altogether,
fractional intensities calculated from intensities of two neighbouring spectral channels
allow to assign the principal emitting unit to a fluorophore of a photonic wire.
Furthermore, the histograms exhibit a certain homogeneity and a poor influence of
photophysical processes, as compared to scan images derived from wires adsorbed
on dry glass substrate.

                                   Results and Discussion

      4-28: Fractional intensities for different detector channels calculated for Wire
      04-01 (A), Wire04-02 (B and C) and Wire 04-03 (D-E). Spectral distributions allow
      to discriminate the emitting species in a photonic wire.

Fractional intensity analysis discussed above for the fluorophores RhG, TMR and
ATTO590 was extended to two more fluorophores, ATTO647 and Cy5.5. Mean
values of distributions are summarized in table 4-21.

                                          Results and Discussion

                            F2 (green, blue)         F2 (yellow, green)   F2 (red, yellow)

     RhG                        0.30                         -                   -

     TMR                        0.67                      0.42                   -

     ATTO590                       -                      0.65                  0.21

     ATTO647                       -                      0.81                  0.33

     Cy5.5                         -                         -                  0.61

Table 4-21: Fractional intensities derived from different combinations of detector channels
characterize five fluorophores contributing to Wire04-10 in aqueous solution.

This method allows to differentiate between spectral emission patterns of single
photonic wire samples observed under aqueous conditions. Nevertheless, a few
comments are necessary for a better understanding. First, a contribution of TMR onto
the blue detector channel cannot be understood from the emission spectrum of TMR,
but is a measure of lower FRET efficiency frequently detected in the FRET-pair RhG-
TMR. This issue was discussed in section 4.1.6 with fractional intensities determined
from fluorescence trajectories of single FRET-pairs. It was found that a value of F2
(green, blue)   = 0.70 equals an energy transfer efficiency of ~80%, which is affirmed by
the spot statistic which is based on data analysis in table 4-21. Similarly, emission
from the last fluorophore Cy5.5 is expected to occur on the “red” detector channel
only. Contrary to this assumption, a value of F2 (red, yellow) = 0.61 was observed. This
fact suggests less efficient energy transfer from the fourth dye to the fifth dye,
ATTO647 to Cy5.5. It is to mention that quantum efficiencies are the main reason for
this observation, which differ in both fluorophores, with values of 0.65 and 0.28
respectively. Nevertheless, the last energy transfer step exhibits a lower transfer
efficiency, but reliable values can not be obtained by this method and will be derived
from time-resolved single-molecule trajectories later.

                                  Results and Discussion


In the last section, the positive effect of changing measurement conditions on single-
molecule    experiments     was    demonstrated.      Inhomogeneous       broadening     of
fluorescence properties are reduced dramatically by measuring in aqueous solution.
A further step towards increased homogeneity and long observation times was made
by working with rhodamine dyes in reductive environment. A glaring example can be
found if single-molecule experiments with RhG performed on dried glass (section
4.3.2) and in aqueous solution (section 4.3.3) are compared.

Besides reducing inhomogeneous broadening of single fluorophores, it is desirable to
have a strategy for a selective construction of multichromophoric molecules, with
reduced chemical heterogeneity. Still, hybridisation of single strands at the ensemble
level yields many subpopulations, as can be seen in the scan images shown in figure
4-27. There are several reasons that can be thought of, starting with concentration
errors, difficulties in hybridising such a complex system out of chemically modified
oligonucleotides, or a chemical equilibrium between different species. To circumvent
these chemical heterogeneities, a method of “forced hybridisation” directly at the
single-molecule level was elaborated.

 Figure 4-29: Sequence of scan images (8 x 8 µm², 2 kW/cm², 50 nm pixelsize and 2 ms/pixel
 integration time) of a surface with Rhodamine Green labelled to ss60bp DNA in PBS/MEA.
 After the first scan image, a 10-8 M solution of a complementary ss20bp DNA labelled with
 TMR and ATTO590 was added. Fluorescence was recorded on three detector channels.

                                    Results and Discussion

In single-molecule hybridisation experiments, a streptavidin coated surface was
treated with a solution of single-stranded DNA labelled with Rhodamine Green at its
5’-end and carrying a biotin anchor at its 3’-end. In a next step, the solution was
washed off and replaced by a second solution, containing the first antisense
sequence of 20bp length with a concentration of ~10-8 M, labelled with both TMR and
ATTO590. By constantly scanning the surface with relatively low excitation power
(which prevents photobleaching of Rhodamine Green dyes), the first binding events
occurred after a few minutes, distinguishable by the appearance of energy transfer.
Since both fluorophores TMR and ATTO590 are attached to the same
oligonucleotide, fluorescence emission is mainly observed from ATTO590, i.e.
detected on the “yellow” detector. A sequence of eight scan 8 x 8 µm² scan images
visualises this process in figure 4-29.

The scan images shown in figure 4-29 were taken every two minutes and show a
considerable process of hybridisation of the antisense oligonucleotide to the single
stranded DNA previously attached to the surface. This process is driven by the
diffusion of the antisense oligonucleotide. Already in the third image, 10 molecules
out of ~60 molecules exhibit photon emission in the “yellow” detector, a tendency
which   continues      in   the   following   images       until   nearly     all   surface-bound
oligonucleotides are hybridised. Eventually, emission is detected on the “blue”
channel, but most molecules exhibit efficient energy transfer from the first
chromophore RhG via TMR to ATTO590. As previously mentioned, hybridisation at
the single-molecule in the way it was demonstrated is a forced process. This can be
visualized by a few numbers. If we assume a typical single-molecule density of one
molecule per square micrometer and observe a volume cube of 10 x 10 x 10 µm3,
this cube contains 100 molecules immobilized at the surface. In the total volume of
the cube (which equals 10-12 l), we will find ~6000 molecules from the antisense
oligonucleotide. These oligonucleotides in the solution need 10 to 100 ms to diffuse
randomly through the volume of the cube. If an appropriate binding site is in range,
the very efficient process of hybridisation can occur, but contrary to equal
concentrations    in    ensemble     hybridisation,    a     possible       interaction   between
oligonucleotides can occur many times until appropriate conditions lead to final

Furthermore, hybridisation of four and five chromophores was demonstrated. Adding
the second single stranded oligonucleotide labelled with ATTO647 showed a similar
                                  Results and Discussion

hybridisation efficiency, as shown in figure 4-30 (C). To better distinguish the
spectrally near emitting fluorophores ATTO590 (see scan image (A) in figure 4-30)

      Figure 4-30: 8x8 µm² scan images of surface-hybridized photonic wires.
      Samples Wire04-03 (A) together with spectrally-resolved emission on “red” and
      “green” channel (B), Wire04-09 (C) with spectrally-resolved emission in (D).
      Five-dyes containing photonic wires shown are Wire04-10 (E), Wire04-11 (F) and
      Wire04-9C (G).

and ATTO647, the spectrally separated images of the “red” and the “green” channel
are presented for each scan image (see (B) and (D) in figure 4-30), showing a higher
contribution on the “red” channel for ATTO647. Less efficiency is observed for the
third antisense oligonucleotide, in dependence of the fluorophore used. Scan images
of Wire04-10 (scan image (E) in figure 4-35, Cy5.5 as final emitter), Wire04-11 ((F),
ATTO725 as final emitter) and Wire04-9C ((G), ATTO680) demonstrate this
observation for different fluorophores. In the case of Wire04-11, the reduced density
of “red” spots (e.g. with respect to Wire 04-10) is explained by a poor photostability of
far-red emitting fluorophores ATTO725 (a similar observation was made for
ATTO740). A relatively high intensity of red-emitting species in the scan image of
Wire04-9C was achieved by incubating a surface with a higher-concentrated solution
(10-7 M) over night. At this point, one must consider the influence of fluorophores on
                                 Results and Discussion

DNA-hybridisation: one of the conspicuous effects of multiple labelling of DNA
oligonucleotides is the degree to which it lowers the melting temperature, Tm. The
source of this destabilization was demonstrated to arise from dye–dye and dye–
nucleotide interactions [Waggoner and Randolph, 1997] (dye-dye interactions were
observed for cyanine dyes in distances of 6 nucleobases and can be neglected for
photonic wires in this work).

In this passage, it could be demonstrated that online DNA-hybridisation of short
oligonucleotide sequences can be exploited for the construction of photonic wires
with up to five fluorophores at the single-molecule level. A higher chemical
homogeneity can be achieved, and a hindered hybridisation step of the third
antisense oligonucleotide is observed.


The mere detection of spectrally-resolved fluorescence of multichromophoric
photonic wire samples relies on spectral changes observed, which consecutively are
assigned to a change of the emitting unit in a photonic wire. In many cases, this
observation is valid, but some scenarios can not be resolved. There can be, on the
one hand, spectral jumps of one chromophore, but also inefficient energy transfer
and fluorescence of two fluorophores simultaneously, which could exhibit a similar
spectral signature as a single fluorophore in the energy cascade. To resolve these
phenomena and contribute to a further characterization of these complex samples,
time-resolved experiments at the single-molecule were realized. Set-ups used for
experiments are described in chapter 3.1.1. Experiments at 476 nm excitation were
performed on dried glass, whereas experiments at 488 nm were performed in
aqueous solution (for surface preparation, see 3.3.2).

Time-Resolved Single-Molecule Experiments of Photonic Wires Adsorbed on
Dry Glass Substrates

For investigation on dried glass substrate, Wire04-06 was chosen, constituted of the
chromophores Rhodamine Green, TMR, ATTO590, LCR and ATTO680 (for

                                    Results and Discussion

       Figure 4-31: Fluorescence trajectory with relative intensities recorded from a
       single photonic wire (Wire04-06). Fluorescence decays for four different parts of
       the trajectory, A-D, are shown below.

fluorescence properties see table 4-9; ATTO620 was replaced by LCR, but exhibits
spectrally similar properties).

An exemplary fluorescence trajectory of this photonic wire sample which was chosen
because of additional information not available in previous experiments is shown in
figure 4-31. For more clarity, relative intensities of each detector channel are added
above the trajectory in figure 4-31, i.e. the fluorescence detected on each channel
normalized to the overall emission. This approach is a little different to fractional
intensities used so far (see 4.3.3), but offers an easier way to determine fluorescence
changes (this method is often used under the name of “proximity values” [Clamme
and Deniz, 2004]). The practical importance can be seen after ~3 s in the trajectory,

                                     Results and Discussion

where a sudden increase in fluorescence is observed, but no spectral change is
observed in relative spectral intensities. The fluorescence trajectory was further
separated into four sections (according to regions which do exhibit different spectral
signatures), named A to D, and fluorescence decays recorded at the different
detector channels are shown in figure 4-31. Fluorescence lifetimes obtained from
decays are listed in table 4-22.

                    “blue”             “green”            “yellow”              “red”

     A                 -            4.0 ns/1.1 ns       4.0 ns/1.1 ns           4.0 ns

     B             < 0.3 ns             1.4 ns          3.2 ns/1.5 ns           3.2 ns

     C                 -                1.0 ns              1.1 ns              1.0 ns

     D              2.8 ns              2.9 ns                 -                   -

Table 4-22: Spectrally-resolved fluorescence lifetimes from the trajectory in figure 4-31.

The mere look on the spectral pattern of fluorescence intensity in the first section (A)
suggests a predominant emission from the fourth dye, LCR (see spectral pattern in
section 4.3.1). Fluorescence lifetime implies another interpretation: a well
pronounced rise time is observed on the “red” detector, and fluorescence on both
“green” and “yellow” channel are strongly quenched (1.1 ns), but still exhibit a
component of 4.0 ns, which is the lifetime of both ATTO590 and LCR. In the next
short period (B), relative intensities change, and the spectral pattern with equal
intensity on the “green” and “yellow” channel could suggest emission of both
ATTO590 and TMR. Again, fluorescence lifetime draws a different picture: a lifetime
of 1.4 ns and 1.5 ns observed for decays of the “green” and “yellow” detectors
suggest an efficient energy transfer from ATTO590 towards the last emitter,
ATTO680. Emission of LCR can be excluded, since a much more pronounced
contribution of 3.2 ns should be observed on the “yellow” detector. Part (C) shows
the typical intensity pattern observed for ATTO590 on dry glass, and lifetime
measurements supply the additional information of a quenched state of this
fluorophore. This could be either due to an absorbing and nonfluorescent
fluorophore, e.g. ATTO680, or due to quenching by guanosine residues neighbouring

                                      Results and Discussion

the dye. The last part (D) shows the emission pattern of green-shifted Rhodamine
Green (see 4.3.2), accompanied by a weak quenching.

Beyond the previous discussion of fluorescence lifetimes observed on different
detector channels, a distributional analysis applying a deconvolution procedure and
using multiexponential fits was made, using custom-made software (LabView,
National Instruments, USA)). The method is very similar to the one described for
time-resolved ensemble-spectroscopy (see section 3.3.2), and uses the IRF of each
separate detector (see 3.1.2). Finally, energy transfer efficiencies for each of the four
steps along the photonic wire can be calculated, and results for parts A-D in figure 4-
31 are summarized in table 4-23.

                      E1                      E2               E3                 E4

     A               0.99                0.99                  0.74              0.43

     B               0.99                0.99                  0.19              0.83

     C               0.99                0.99                   0                 0

     D                 0                      -                 -                  -

Table 4-23: Energy transfer efficiencies for each of the four transfer steps of different parts of
the fluorescence trajectory in figure 4-31.

Values listed in table 4-23 complete the picture drawn in a first description: in part (A)
of the trajectory, the first two steps exhibit highest energy transfer efficiencies,
whereas values decrease for the third and fourth step. In (B), no efficient transfer
towards LCR is observed, whereas still strong energy transfer towards ATTO680
occurs (at this point, it can be assumed that LCR is photobleached, and that the
fourth rate describes the energy transfer from the third dye to the last one). Part (C)
shows emission of ATTO590 predominantly, although a strong quenching is
observed, and part (D) shows no contribution to any energy transfer rate, i.e.
Rhodamine Green is finally the single emitter.

To verify lifetime calculations and interpretations made above, the relative
contributions of each fluorophore can be summed up to obtain a spectral pattern,
which again can be compared to the spectral intensity pattern observed in the
trajectory (see table 4-24).

                                     Results and Discussion

                    “blue”             “green”             “yellow”              “red”

    A              0.02/0.06           0.18/0.18           0.47/0.41           0.32/0.36

    B              0.02/0.02           0.40/0.29           0.42/0.46           0.16/0.23

    C              0.03 / 0.04         0.22/0.32           0.61/0.51           0.13/0.13

    D              0.55/0.58           0.36/0.37           0.07/0.05           0.02/0.01

Table 4-24: Relative intensities for each detector for parts A-D of the fluorescence trajectory in
figure 4-31, observed/calculated values.

Values in table 4-24 represent one linear combination of single fluorophores which
was obtained by pure lifetime analysis. A good accordance to measured intensities
can be stated from these results.

The important information obtained in this section can be summarized as follows: a
constant spectral pattern of a single photonic wire observed cannot necessarily be
explained by only one linear combination of spectral patterns from single
fluorophores. Insufficient energy transfer steps cannot be resolved, and a “leak” in
the chain of transmitting units (as observed for LCR in the previous discussion) may
lead to false interpretations. Time-resolved analysis can unravel fluorescence data
and contribute to an improved understanding of the functionality of such complex
optical devices.

Time-Resolved Single-Molecule Experiments of Photonic Wires Immobilized in

Similar experiments were performed with photonic wire samples immobilized in
solution (see 4.3.3) using a mode-locked titan sapphire laser, operating at 488 nm
with a repetition rate of 80.77 MHz (see 3.1; the identical set of filters and
beamsplitters was used as for continuous wave excitation, see 4.3.1).

A fluorescence trajectory of Wire04-10 is presented in figure 4-32. The sample is
constituted of the chromophores Rhodamine Green, TMR, ATTO590, ATTO647 and
Cy5.5, which allows a better identification of the last fluorophore by fluorescence
lifetime (a value of ~1 ns is observed for the free dye, whereas bound to DNA, the

                                    Results and Discussion

4-32: Fluorescence trajectory of a single Wire04-10. The trajectory exhibits a signature of
sequential photobleaching from the last acceptor Cy5.5 over ATTO647, ATTO590, TMR and
finally Rhodamine Green. Decays for five representative regions are shown and confirm the
interpretation derived from spectral signatures of fluorescent dyes.

fluorescence lifetime increases to ~1.5 ns). For stability reasons, single-molecule
measurements were performed in PBS/MEA (see 4.1.2-4.1.5; rhodamine derivatives
and carborhodamines yield higher photon counts, and carbocyanine dyes are not
affected in their fluorescence properties if observed in reductive environment).

The fluorescence trajectory represents an example for a rarely observed (and also
not expected event) of sequential photobleaching of a photonic wire: five well-
distinguishable levels of emission can be discriminated already by a mere look on the
spectral characteristics of emission. The first 1.1 s of the trajectory show emission on
the red channel. This observation is supported by a significant rise time appearing on
the “red” channel, together with a fluorescence lifetime of 2.1 ns. A slightly longer
lifetime of 2.3 ns is observed on the “yellow” channel, indicating that this step only
transfers a part of the energy. Both “blue” and “green” channel are characterized by
strong quenching caused by energy transfer, exhibiting a lifetime of ~1.2 ns. In the

                                      Results and Discussion

second step, fluorescence emission is dominated by ATTO647 (4.2 ns for both
“yellow” and “red”). Due to a higher quantum yield of ATTO647 (~0.65) compared to
the previous emitter Cy5.5 (0.28), an increase in total intensity is observed. Between
~2 s and ~3 s, a collective “off” period is observed on the “yellow” and “green”
channel, but still photons are detected on the “blue” and “green” detector (1.3 ns).
Therefore, a fluorescent trap of either ATTO647 or ATTO590 can be postulated, still
absorbing but not exhibiting fluorescence anymore. Fluorescence is recovered at
~3 s with the spectral and temporal signature of ATTO590, followed at ~6 s by TMR
and at ~9 s by Rhodamine Green. Interestingly, fluorescence emission is still
observed after ~13 s for Rhodamine Green with a similar fluorescence lifetime of ~4
ns. A summary of fluorescence lifetimes of various parts of the trajectory in figure 4-
32 is given in table 4-25.

Time / s                     “blue”         “green”            “yellow”          “red”

0.00-1.05                    1.4 ns          1.2 ns             2.3 ns          2.1 ns

1.05-2.10                    1.3 ns          1.3 ns             4.2 ns          4.2 ns

2.10-3.00                    1.3 ns          1.3 ns               -                -

3.00-6.50                    1.5 ns          2.3 ns             4.4 ns          4.2 ns

6.50-9.00                    1.8 ns          3.9 ns             4.1 ns             -

9.00-12.25                   4.2 ns          3.4 ns               -                -

12.25-end                    4.0 ns          3.8 ns

Table 4-25: Fluorescence lifetimes for each detector channel determined from the fluorescence
trajectory in figure 4-32.

Due to a broader instrument response function exhibiting a “tail” (see section 3.1.1),
the method of deconvolution could not be used. Mathematical fits did not lead to a
satisfactory convergence, and energy transfer efficiencies could not be determined in
a similar way. Nevertheless, the information won at this point is the occurrence of
collective “off” states, whereas still quenched emission from the first (and eventually
second?) fluorophore is observed. Energy transfer efficiencies with a carbocyanine
dye as last emitter of a photonic wire make it possible to determine energy transfer
efficiencies easier due to a shorter lifetime. This is especially important for the last
                                  Results and Discussion

energy transfer step, which has proven to be the weakest in many experiments
(compare to paragraph 4.3.2, 4.3.3, and the beginning of this section). In the
exemplary trajectory of figure 4-32, the transfer efficiency of the last step can be
estimated by the change in lifetime of ATTO647, serving as the donor for Cy5.5, and
yields a value of E4 = 1 – (2.3 ns/4.2 ns) = 0.45. Furthermore, more than one intensity
level (but equal spectral patterns) are observed for Rhodamine Green, which merit a
more detailed analysis of this fluorophore by an extended technique in the following


If we think about the structural organization of a photonic wire sample, many possible
configurations and conformations might be possible. These are of great importance
for the energy transfer method chosen for the design of these compounds in this
work, i.e FRET, since the absolute orientation and the mobility of a fluorophore has a
major influence on transfer efficiencies (see 2.1.2). The ideal case of κ2 = 2/3 for
randomly orientated and diffusing molecules might therefore be an assumption which
fails in some cases.

To elucidate the structural mobility of Rhodamine Green, the polarization of the 488
nm excitation light source (TiSaph, see 3.1 and 3.1.1) was modulated regularly using
an electrooptical modulator (EOM, Linos, USA). Photon absorption occurs
preferentially if the orientation of the electric field vector of light is parallel to the
absorption dipole moment of the chromophoric unit of a fluorophore (see 1.1.1). If a
fluorophore is freely rotating, no modulation should be observed, whereas a immobile
fluorophore should show modulation in fluorescence. Experiments were carried out
with Rhodamine Green bound to DNA and immobilized in water (see 4.1.5).
Excitation light modulation was done with a frequency of 20 Hz. Fluorescence
trajectories from single Rhodamine Green molecules are shown in figure 4-33.

In part (A) of figure 4-33, a fluorescence trajectory with two steps in intensity is
shown, recorded from one single fluorophore. A magnified view at the transition
between both intensity levels portrays the rotational mobility. In both parts of the
trajectory, fluorescence emission follows the modulation of the polarization, which
indicates hindered rotation of the molecule. Two different fluorescence lifetimes are
                                    Results and Discussion

    Figure 4-33: Single-molecule trajectories of two Rhodamine Green molecules derived
    by modulation of the polarization of the excitation light. Both molecules exhibit
    different degrees of rotational mobility. Fluorescence decays of assigned parts a-d of
    the trajectories are shown for each fluorophore.

observed, i.e. 3.4 ns in the part a (0 – 4.2 s) and a strongly-quenched lifetime of
below 1 ns in part b (which cannot be resolved further due to pronounced broadening
of the IRF, see 3.1.1 and 4.3.5).

                                   Results and Discussion

Different to the first trajectory, the second one shown in part (B) of figure 4-33 shows
many levels of rotational mobility in one single fluorophore: during the first 0.5 s of
observation time, the fluorophore exhibits fluctuations in fluorescence intensity
according to the modulation frequency of the excitation light. This is followed by a
period of 0.5 s which exhibits nearly no modulation, indicating a freely rotating dye.
Fluorescence lifetime in both cases is very similar, values of 4.1 ns (part c) and 4.2
ns (part d) are determined. From a time of 1 s on, an even stronger modulation of
fluorescence is observed until the fluorophore goes into a state with substantially
lowered quantum yield of fluorescence (~2.1 s).

A similar observation of changing fluorescence lifetimes and quantum yields was
made for the dye TMR [Eggeling et al., 1998b]. Accordingly, three different states for
Rhodamine Green bound to DNA can be postulated, which are:

              (1)   A freely-rotating fluorophore (no modulation, lifetime of 4.2 ns)

              (2)   Hindered rotation and high quantum yield (modulation of
                    fluorescence, but similar lifetime of ~4 ns)

              (3)   Hindered rotation and strongly quenched intensity (modulation of
                    fluorescence and quenched lifetime down to ~1 ns)

States with hindered rotation can be attributed to interactions of the fluorophore with
nucleobases (state (2)) and possible quenching reactions, e.g. with the guanosine
residue at position 4 (state (3), compare to figure 4-14 in section 4.2.1 for the DNA
sequence of the single-stranded 60bp oligonucleotide).

These different structural configurations observed for Rhodamine Green complicate
the understanding of the working principle of photonic wires. The orientation of a
fluorophore which is part of an energy transfer cascade is crucial for transfer
efficiencies, and certainly the assumption of freely rotating fluorophores cannot be
held. On the other hand, single-molecule experiments can refine the understanding
from steady-state experiments (section 4.2.2 and 4.2.3) and explain lower energy
transfer efficiencies observed for the first step in photonic wires.

                                 Results and Discussion

4.4. Carbocyanine Dyes as Optical Single-Molecule Switch

Single-molecule fluorescence experiments have revealed several expected and
unexpected photophysical phenomena of the carbocyanine dye Cy5 such as cis-
trans isomerisation, “off”-states additional to triplet formation, and complex
photobleaching pathways including nonfluorescent intermediates that still absorb light
in the visible range [Widengren and Schwille, 2000; Ha et al., 1999; Tinnefeld et al.,
2003; Ha and Xu, 2003]. Beyond the characterisation of fluorescence intermittencies
observed for Cy5 which were presented earlier (see section 4.1.2), the photophysical
characteristics of this fluorophore represent a pivot point for extended studies. At
least three intermediate states towards photobleaching were discovered [Ha and Xu,
2003], and it was one focus of this work to further investigate the complex
photodestruction pathway and to influence it by optical means. Transitions into such
intermediate states offer a possibility to use conventional fluorophores as single
molecule switches, presumed that the transition is reversible.

In general, controlled on/off switching of the fluorescence of a single chromophore at
room temperature affords the introduction of a controllable and highly efficient
competing quenching pathway that prevents emission from the excited singlet state
of the chromophore via, for example, excitation energy transfer or photoinduced
electron transfer. Switching might be accomplished by light-induced deactivation of
the quencher or changes in the chromophore/quencher interaction geometry [Liang
et al., 2003]. More recently, the first room temperature single-molecule photoswitch
based on optical switching of the transfer efficiency in a FRET-pair was published
[Irie et al., 2002; Fukaminato et al., 2004]. In a two-colour experiment, it was
demonstrated that a donor chromophore (bis(phenylethynyl)anthracene) connected
to a switchable quenching unit (a diarylethene derivative) could be switched on and
off by 488- and 325-nm light, respectively. UV light was used to activate the
quencher (energy transfer acceptor), while 488-nm light was used for deactivation of
the quenching unit and probing of the fluorescence of the donor chromophore. The
use of identical wavelengths (488 nm) for probing and switching was possible
because the deactivation (isomerisation) is about 1000 times less efficient than the

                                  Results and Discussion

activation of the quenching unit. Thus, probing and isomerisation can be controlled
by changing the excitation light intensity.


To probe intermediate states of individual Cy5 molecules, oxygen has to be removed
efficiently, since irreversible photooxidation would be the consequence. As presented
in section 4.1.2, the observation time of an individual carbocyanine prior to
photodestruction is relatively short and makes further studies of intermediate states
impossible.   To   eventually    change       the   chemical   environment   and   reduce
inhomogeneous broadening, carbocyanine derivatives Cy5 or Alexa 647 (Molecular
Probes, USA) were thus coupled covalently to double-stranded biotinylated DNA and
immobilized on surface under aqueous conditions. A dual-laser set-up with two
excitation wavelengths was used for two-colour excitation of single molecules, i.e.
488 nm and 632.8 nm supplied by an argon ion laser and a helium-neon laser,
respectively (see 3.1.3). Fluorescence light was split onto two detectors using a
dichroic beamsplitter (680DRLP) and two bandpass filters (675DF50, 700DF75).

A series of scan images recorded from individual Cy5-molecules in PBS with 100 mM
MEA is shown in figure 4-34.

First, a fluorescence image was recorded exciting the sample at 633 nm to verify the
presence of single immobilized labeled dsDNA molecules (figure 4-34, A). Depending
on the excitation intensity, most molecules were found in a nonfluorescent state after

 Figure 4-34: Series of fluorescence scan images demonstrating the principle of optical
 switching of Cy5-molecules. Image A was recorded at 633 nm and led to an important
 reduction of fluorescent molecules B. After scanning the same surface with 488 nm,
 fluorescence of most molecules is recovered, demonstrated in the scan image C.

                                   Results and Discussion

the first image scan (figure 4-34, B). After the same area was scanned with 488-nm
laser light under otherwise identical conditions, the fluorescent state was recovered,
as shown in figure 4-34, C.

Selection of a single Cy5 molecule in an image scan and monitoring its fluorescence
with time under alternating or simultaneous 488- and 633-nm excitation enables the
investigation of the switching behaviour in more detail. Figure 4-35 A shows a
fluorescence intensity trajectory of a single Cy5 labelled dsDNA molecule
continuously excited at 633 nm. After about 500 ms, the fluorescence ceased and did
not recover for 5 s of 633-nm laser irradiation. The bright state could, however, be
reproducibly recovered within a few hundred milliseconds by simultaneous irradiation
at 488 nm (after 5 s in the trajectory of figure 4-35, A). Subsequently, driven by
simultaneous irradiation at 488 and 633 nm, the molecule switches between a

 Figure 4-35: (A) Fluorescence trajectory of a single Cy5-labeled dsDNA molecule in
 deaerated PBS, pH 7.4. During the first 5 s, the sample was irradiated at 633 nm with an
 excitation intensity of 14 kW/cm2. After ~500 ms, the fluorescence of the molecule
 disappeared and did not recover. At 5 s, the molecule was irradiated simultaneously at 488
 nm with equal intensity (14 kW/cm2). Subsequently, the fluorescence recovered and ceased
 in an alternating fashion. (B) Reversible optical switching of a single Cy5 molecule. The
 molecule was irradiated at 633 nm until fluorescence ceased (generally within 1 to 2 s) and
 then recovered by irradiation at 488 nm for 2.5 s. The underlayed colour indicates the
 excitation wavelength (blue: 488 nm, red: 633 nm).

                                 Results and Discussion

fluorescent and nonfluorescent state on a time scale of a few hundred milliseconds.

To demonstrate the applicability of single Cy5 molecules as controllable and
reversible optical switches, a single Cy5-labeled dsDNA molecule was irradiated at
633 nm until fluorescence ceased (generally within 1 to 2 s) and then recovered by
irradiation at 488 nm for 2.5 s (trajectory (B) in figure 4-35). Out of 21 on/off cycles
shown, the fluorescent state of Cy5 could be recovered 20 times upon irradiation at
488 nm with an excitation intensity of 14 kW/cm2. Switching failed only once either
because irradiation at 488 nm was not sufficient to recover the emissive state or
because the time the molecule spent in the fluorescent state was too short to detect
a sufficient number of fluorescence photons for an unequivocal discrimination against
background signal.


The efficiency of Cy5 as a photoswitch strongly depends on the buffer conditions
used: For best performance, oxygen has to be removed rigorously and a triplet
quencher such as MEA has to be added. The need for a triplet quencher also
indicates that the triplet state is not involved in the formation of the nonfluorescent
switchable state. Under such conditions, more than 100 switching cycles could be
achieved for single Cy5 molecules with a reliability of >90%. While the efficiency of
switching strongly depends on switching conditions (i.e., oxygen, MEA, and
irradiation wavelength), it does not require double-stranded DNA. Both biotinylated
Cy5 and Cy5-labeled single-stranded DNA immobilized on BSA/ biotin-streptavidin
coated glass substrates could be switched as well with comparable efficiency
irradiating either at 488 or at 532 nm. Under dry conditions in the presence of
oxygen, however (e.g., adsorbed on bare glass surface), the fluorescent state could
not be recovered. Furthermore, switching could also be observed for immobilized
Alexa647, a structural related dye which implies that the underlying mechanism
constitutes a general feature of certain carbocyanine dyes.

To investigate the nature of the switchable state and the switching mechanism, the
photophysics of Cy5 under 633-nm irradiation was compared to the photophysics of
Cy5 simultaneously using 633 and 488 nm irradiation. In agreement with Widengren
and Schwille and previously presented results for Cy5, autocorrelation of

                                   Results and Discussion

 Figure 4-36: (A) Fluorescence intensity trajectories of single immobilized Cy5-labeled DNA
 molecules show additional off states with durations of several milliseconds (oxygen-free
 PBS without triplet quencher, excitation intensity of 3 kW/cm2 at 633 nm). (B)
 Autocorrelation of an intensity trajectory recorded in the absence of a triplet quencher
 shows “off” states with three distinct time scales: cis-trans isomerisation (iso), triplet
 states (isc) at reduced oxygen concentration, and an additional off state in the millisecond
 range. Additionally, a three-exponential fit plus “on” and “off” times for three independent
 processes is given. (C) Trajectory of a single Cy5-labeled DNA molecule under
 simultaneous 488- and 633-nm excitation (each with 3 kW/cm2), 100 mM of the triplet
 quencher MEA was added. (D) Off-times histogram of Cy5 trajectories as in (C). (E) In the
 absence of a triplet quencher and oxygen, Cy5 exhibits dynamic changes in fluorescence
 intensity and blinking parameters (excitation at 633 nm with 14 kW/cm2).

fluorescence intensity trajectories of immobilized Cy5 molecules revealed two
components, which can be ascribed to intersystem crossing and cis-trans

                                 Results and Discussion

isomerization [Widengren and Schwille, 2000] In these experiments, oxygen was
removed while no triplet quencher was added to separate the time scale of triplet
blinking and cis-trans isomerisation. Additionally, another off state in the lower
millisecond time range (2-20 ms) is visible in the fluorescence intensity trajectory
(figure 4-36, A) as well as in the autocorrelation function of immobilized molecules
(figure 4-36, B). In the presence of triplet quencher (100 mM MEA) and oxygen
scavenger, simultaneous irradiation at 488 and 633 nm produces similar intensity
fluctuations and an additional “off” state with a duration of about 200 ms (figure 4-36,
C). The additional “off” state in the higher millisecond time scale only appears with
simultaneous 488-nm excitation. As a consequence, this long “off” state can be
assigned to the photoswitched state. As these rare long “off” times are not well
amenable to autocorrelation analysis, “off”-time histograms (figure 4-36, D) were
used for closer analysis. The first bins in the histogram of figure 4-36 (D) represent
predominantly the shorter millisecond off state. Longer off states show a broad
distribution on time scales of up to several seconds. In contrast to the “on” state,
which shows a simple power dependence (i.e., independent of 633-nm excitation
power), on average 5000 photons could be detected during each “on” time), and the
long “off” times show a weak 488-nm excitation power dependence (figure 4-36, D),
indicating the presence of an additional intermediate that might be formed thermally.
However, the broad distribution of “off” states and also significant differences in the
time scale of this “off” state from molecule to molecule suggest that the local
environment of the chromophore could play an important role. Although the
molecules are immobilized in a way that surface interactions are minimized,
temporary interactions of the chromophore with the DNA and proteins used for
immobilization cannot be excluded. Evidence that changes in the local environment
influence the photophysics of single Cy5 molecules comes from the observation of
triplet blinking fluctuations (figure 4-36, E). In the absence of triplet quencher, the
observed changes in blinking pattern account for variations in intersystem crossing
yield and triplet lifetime, which are caused by varying local oxygen concentration,
different interaction geometries with DNA or protein, or other influences of the
macromolecules on the chromophore [Köhn et al., 2001].

To further characterise the photophysical properties of the reversible switchable state
and to evaluate the implications of photoswitching for common single-molecule FRET
experiments, TMR as an energy transfer donor was attached, together with Cy5 to

                                       Results and Discussion

 Figure 4:37: (A,B) Fluorescence trajectories of donor-acceptor (TMR-(N)10-Cy5)-labelled
 dsDNA in PBS/MEA. (A) After preparation of Cy5 in its nonemissive state (irradiation at 633
 nm for 2 s), the molecule was excited at 514 (14 kW/cm2) starting at 3 s to probe donor
 emission. Appearance of the donor emission demonstrates that the nonfluorescent state
 does not quench donor emission. After a few hundred milliseconds, the fluorescent state of
 the acceptor recovers, thus quenching the donor emission via resonance energy transfer.
 As expected from the other experiments, Cy5 undergoes several additional transitions to
 nonfluorescent states during the experiment. The underlayed colour indicates the excitation
 wavelength (light blue: 514 nm, red: 633 nm). The expanded view of a switching event
 uncovers the presence of an additional nonfluorescent state of Cy5, which efficiently
 quenches the donor emission (indicated by τdelay). (B) The arrows indicate Cy5 off states in
 the lower millisecond time scale that quench TMR more efficiently than the fluorescent
 state. (C) Lifetime histogram of τdelay.

dsDNA. For a separation of ~3.4 nm (10 base pairs), strong coupling of donor and
acceptor or the formation of ground-state complexes is prevented, and the donor
transfers its excited-state energy via nonradiative dipole-dipole interaction efficiently
to the acceptor Cy5. Therefore, the donor fluorescence can be advantageously used
to report on the photophysical states of the acceptor Cy5. Figure 4-37 (A) shows a
fluorescence trajectory recorded from a single double-labelled DNA molecule
recorded in standard buffer. During the first few seconds of the experiment, Cy5 was

                                   Results and Discussion

“prepared” in its nonfluorescent state by 633-nm excitation. Upon changing the
excitation wavelength from 633 to 514 nm after ~3 s, the donor fluorophore TMR
shows strong emission, demonstrating that the absorption of the nonfluorescent state
of Cy5 is not in resonance with TMR emission (i.e., it does not absorb in the spectral
range of the donor emission between 570 and 650 nm). Hence, no energy transfer
occurs. A few hundred milliseconds later, however, the fluorescent state is recovered
by 514-nm excitation and energy transfer takes place with high efficiency until Cy5
enters again the nonfluorescent state. Subsequently, Cy5 switches between the
nonfluorescent and fluorescent states. Closer examination of the switching events
shown in figure 4-37 (A) reveals that TMR is nonfluorescent about to the recovery of
the fluorescent Cy5 state. Although this off state, which is denoted τdelay, covers the
same time scale as frequent donor blinking (see Figures 4-36 (B) and 4-37 (A)),
statistical analysis suggests that the probability of Cy5 recovery is much higher when
TMR is not emitting. Thus, one can assume that τdelay is due to quenching by the
nearby Cy5 residing in an additional “off” state. The distribution of τdelay is exponential
with a decay time of 2.9 ms (figure 4-37 (C)). Because of the 3.4-nm separation,
quenching of TMR by Cy5 in this nonfluorescent state occurs most likely via
resonance energy transfer. This implies that the absorption of the state is in
resonance with the emission of the donor. Furthermore, the efficiency of energy
transfer from TMR to Cy5 in the quenching state appears to be more efficient than
the standard energy transfer from TMR to Cy5 in its fluorescent state (see donor
intensity in figure 4-37 (B) or in the expanded view of figure 4-37 (A)). Therefore, we
conclude that the short-lived nonfluorescent state of Cy5 exhibits a slightly higher
absorption cross section at shorter wavelengths. It is also possible that this off state
is related to the millisecond off state of Cy5 discussed above (figure 4-36 (A)). As
shown in figure 4-37 (B), this off state is also capable of completely quenching the
donor emission. The fact that it exhibits a similar lifetime and similar acceptor
properties could mean that these two off states represent the same intermediate.

Overall, this study indicates that there are several nonfluorescent states involved in
the switching cycle. The oxygen and triplet dependence of the switching performance
propose that the triplet state is not involved in the formation of the switchable states
but, rather, competes with it. After formation of the first photoswitched product there
seems to be a cascade of events to restore the fluorescent state. “Off” times together
with FRET trajectories suggest one photoactive intermediate and two further
                                   Results and Discussion

intermediates, one with a duration in the 200-ms range and another with a lifetime of
a few milliseconds. The 200-ms state does not function as FRET acceptor, whereas
the state with a lifetime of a few milliseconds is able to absorb the donor energy more
efficient than fluorescent Cy5. Furthermore, this short-lived nonfluorescent state
exhibits similar properties as an off state also found for Cy5 using 633-nm excitation
only and might therefore represent the same state.

One known state of Cy5 that likely exhibits an R0 value larger than that in Cy5 in the
fluorescent state is the cis-conformation of Cy5 [Widengren and Schwille, 2000;
Tinnefeld et al., 2003]. Back-isomerization from the nonfluorescent cis to the
fluorescent trans conformation also occurs photoinduced. Because of the low rates
for thermal relaxation of the photoinduced cis state, photostationary equilibrium is
established between the two isomeric forms already at low excitation intensities. That
is, under laser irradiation a single Cy5 molecule stays in its nonfluorescent cis state
about 50% of the time. trans-cis and back-isomerization generally occur on the
microsecond time scale for Cy5 in solution (figure 4-36 (B)). On the other hand, it can
be anticipated that Cy5 molecules in free solution behave different from those
attached to biomolecules and immobilized onto solid supports. If Cy5 is attached to
DNA or proteins the conformational flexibility of the linker used enables occasional
sticking of the dye on the DNA or protein [Eggeling et al., 1998b]. The steady-state
fluorescence anisotropy of Cy5 attached to dsDNA was determined to be r = 0.24 in
ensemble measurements (i.e., it cannot be regarded as free rotor). Thus, the cis
state could be stabilized for micro- to milliseconds. Accordingly, the off state with a
duration of a few milliseconds that functions as an efficient FRET acceptor could be
associated with a stabilized cis-Cy5 adsorbed to surrounding macromolecules. The
duration of the off state would consequently represent the time of the sticking event.
Even though the triplet state is not involved in the formation of the nonfluorescent
state(s), it can still play a role in the restoration of the fluorescent state. For example,
intersystem crossing of the nonfluorescent state into the triplet state and subsequent
conversion to the triplet state of fluorescent Cy5 followed by intersystem crossing to
the singlet ground state might also contribute to the switching mechanism.

                                   Results and Discussion


In addition, optical switching does not require single-molecule conditions, that is, high
excitation intensity. The fluorescent state can be reproducibly but only partly restored
as well from an ensemble of molecules irradiating, for example, a 10-6 M aqueous
Cy5 solution first at 647 nm and then, to restore the fluorescent state, at either 337,
488, or 532 nm (figure 4-38).

  Figure 4-38: Ensemble switching experiment of an argon-bubbled 10-6 M aqueous Cy5
  solution (PBS, pH 7.4, containing 100 mM MEA). The original Cy5 absorption spectrum is
  shown in black. After being bleached for 30 min at 647-nm (300 mW) irradiation by a
  defocused laser beam, the absorption decreased by ~50% at 650 nm (red). About 40% of
  the absorption could be restored upon irradiation at 488 nm (300 mW) for 30 min (blue).
  The inset shows the temporal evolution of bleaching and absorption restoration. While
  the bleaching curve has to be fitted with a two-exponential model with decay times of τ1 =
  3.1 min (32%) and τ2 = 162 min (68%), absorption recovery is well-described by a single
  exponential with τ = 23 min.

Most notably in the ensemble experiment, the extinction of the reversible off state
shows an increased absorption around 310 nm. Although no substantial increase in
extinction is observed in the range of ~450-532 nm, it is advantageous to use this
wavelength range for switching as the fluorescent subpopulation of Cy 5 is not

                                  Results and Discussion

excited significantly. A small band appearing at ~500 nm after continuous bleaching
is not related to the photoswitched product as the band does not vanish when Cy5 is
switched on again. It is also interesting to note that switching is more efficient at the
level of single molecules than at the ensemble level. In ensemble experiments on
average only about 40% of the molecules can be switched on independent of the
irradiation wavelength. The different switching efficiency observed at the single-
molecule and ensemble levels might be caused by the different experimental
conditions such as different oxygen-removing efficiencies, different excitation
intensities, and/or different local environment on a surface and in solution.

5. Conclusion and Outlook

In this work, an DNA based unidirectional photonic wire and a molecular photoswitch
were realized. Both nanooptical devices can be selectively addressed by light, which
circumvents the connection problem to macroscopic objects. The working principle
and performance of the two molecular elements was investigated by steady-state
optical methods, time-resolved ensemble spectroscopy and single-molecule
fluorescence techniques. In particular, experiments carried out at the single-molecule
level revealed information about important details of the working principle. For
individual photonic wire molecules, energy leaks were identified, the influence of
rotational mobility was investigated and collective nonfluorescent states were
observed. Photoswitches were realized with single carbocyanine dyes. These
fluorophores exhibit a number of intermediate states which strongly depend on
environmental conditions. Two-colour experiments showed efficient and reversible
transitions from the fluorescent state to a dark state. Single-molecule FRET-
experiments with a well-suited donor fluorophore probing the dark state of a
carbocyanine yielded further kinetic information on intermediate states and helped to
elucidate their absorption properties.

It is the intent of the following section to draw a picture of the working principle of
photonic wires and optical switches, as it is derived from the present work. Pivot
points for further studies are discussed which would lead to a more precise
description of the working principle. Finally, possible applications of these
nanooptical devices are introduced.

5.1. DNA Based Photonic Wires

First studies on photonic wires were published in 1994 and used the porphyrin-
approach [Wagner and Lindsey, 1994]. A boron-dipyrromethene dye provided an
input unit at one end, a linear array of three zinc porphyrins were employed as signal
transmission elements, and a free base porphyrin provided an optical output unit at
the other end. Total energy transfer efficiencies of up to ~70% were observed in
these molecular constructs with a total length of ~9 nm.

                                  Conclusion and Outlook

In this work, a stepwise design of a molecular photonic wire using DNA as rigid
scaffold and conventional fluorophores as energy-transferring units was presented.
The advantage of the approach lies in a simplified synthesis strategy based on DNA
as rigid scaffold, and the large variety of fluorophores available for conjugation to
DNA. Sequential hybridisation enabled the construction of multi-dye labelled photonic
wires with appropriate distances for energy transfer in the FRET regime.

To outline the working principle of the photophysical and structural properties of such
multistep energy transfer systems, a number of different techniques were used.
Among steady-state and time-resolved ensemble experiments, single-molecule
experiments especially contributed to a refined understanding of synthetic photonic
wires. A pictorial view of an exemplary photonic wire emphasizing several working
points critical for the development and further improvement of the performance of
such devices is presented in figure 5-1.

    Figure 5-1: Model construct of a photonic wire. Focus was set on the rotational
    mobility of fluorophores, energy transfer efficiencies in single steps, search for
    leakages by time-resolved measurements and evaluation of hybridisation steps.

Rhodamine Green as Input Unit

Rhodamine Green was chosen as input unit for all photonic wire samples
synthesized in this work. As a rhodamine derivative, the fluorophore exhibits a high
number of emitted photons in a         reductive environment (see 4.1.5) and shows
improved stability compared to other short-wavelength absorbing fluorophores.

                                 Conclusion and Outlook

Steady-state and single-molecule fluorescence experiments of this input unit were
one main focus of this work, since an efficient and funnel-like “collection” of light is
crucial for the following processes of energy transfer along the multichromophoric

First, steady-state measurements probed the efficiency of energy transfer exciting
Rhodamine Green in the presence of an acceptor dye, TMR. These experiments only
exhibited poor energy transfer efficiency with a mean value of 0.47 (see 4.2.3), which
is low when compared to the theoretically expected efficiency of 0.99. Time-resolved
spectroscopy revealed two populations with different FRET efficiencies: one with very
high FRET efficiency (0.83 for 23% of the molecules) and a second with a very low
efficiency (0.13 for 77% of the molecules). The second subpopulation was
determined from a fluorescence lifetime of 3.66 ns which is significantly different from
the value observed for Rhodamine Green only (but bound to dsDNA), i.e. 4.20 ns.
Remarkably, a similar observation was made for Rhodamine Green and Cy5 in direct
proximity: ensemble studies revealed a proximity value of 0.44, whereas single-
molecule experiments revealed two equal distributions, one at zero and another at
0.95 [White et al., 2004].

An explanation for the heterogeneity observed for the FRET-pair Rhodamine
Green/TMR can be given by a closer look at the underlying energy transfer
mechanism. Here, the orientation of fluorophores is a crucial factor, which leads to
the interpretation that one subpopulation might be unfavourable for energy transfer.
Orientation effects are taken into account by the κ2 factor, for which a value of 2/3 is
assumed in the case of freely-rotating molecules. In many cases, this assumption is
not totally correct, and theoretical work on orientation probabilities showed a
pronounced contribution at values around zero for the κ2 factor, which might explain
poor transfer efficiencies [Dale and Eisinger, 1979]. On the other hand, single-
molecule studies on the rhodamine derivative TMR bound to DNA revealed three
subpopulations with different fluorescence lifetime and quantum yield, the result of an
interaction with nucleobases [Eggeling et al., 1998b]. This observation might hint at a
similar behaviour of Rhodamine Green and could explain a subpopulation which
does not show energy transfer (or only weak energy transfer). To evaluate this
assumption, experiments with modulated polarization of a pulsed excitation light
source were performed, in order to characterize the mobility of the fluorophore

                                  Conclusion and Outlook

simultaneously with its fluorescence lifetime (see 4.3.6). It can be assumed that a 60
base pair double-stranded DNA does not rotate on the timescale chosen for
modulation (20 Hz), and hence all rotation is attributed to the fluorophore itself.
Structurally different states were observed for one molecule, exhibiting distinct
mobility and fluorescence lifetimes. In many cases, however, hindered rotation was
observed, and interactions with nucleobases may therefore lead to conformations
unfavourable for efficient energy transfer (similar to TMR).

An extension to experiments with modulated polarization of excitation light could be
polarized detection. This would allow the detection of the polarization of fluorescence
emission of each fluorophore in the multichromophoric chain, and would yield
additional information on the working principle of photonic wires. To maintain the
spectral resolution of the present set-up, four additional detectors are necessary,
together with polarizing beamsplitters for each detector channel. Such an extension
of the set-up would require a change in the geometry of the detection path and is
probably difficult to realize with the present 2f-scheme. More flexibility for extension is
envisable if the detection path is designed for parallel light which is finally is focussed
onto each detector separately.

On the other hand, no such subpopulations were observed when hybridisation was
carried out directly on surfaces. A nearly perfect homogeneity of smFRET-
efficiencies could be realized. This observation is contrary to assumptions made
above, but would better correspond to a general heterogeneity in hybridisation at the
ensemble level.

Energy Transfer Efficiencies and Hybridisation

The method for the design of molecular photonic wires which was chosen in the
present work makes use of the efficient process of DNA hybridisation of up to four
single-stranded oligonucleotides. As a result, up to five fluorophores constitute an
energy-transferring chain (see 4.2.1). It was shown that “classical” hybridisation of
single-stranded DNA yields many products besides the desired construct. This large
heterogeneity is also observed in single-molecule experiments, both on dry glass
substrates and immobilized in aqueous environment. Among a number of reasons
that can be given are uncertainties in chemical stoichiometry and possible

                                 Conclusion and Outlook

unfavourable conformations of oligonucleotides which represent a barrier for efficient

A strategy to rule out these heterogeneities was presented in this work (see 4.3.4).
Sequential hybridisation of oligonucleotides to a template single strand bound to a
surface yielded ~90% of desired constructs. The hypothesis of structural hindrance at
the last hybridisation step was verified by longer incubation times and higher

The big advantage of sequential hybridisation lies in the fact that energy transfer
efficiencies of photonic wires can be estimated stepwise at the single-molecule level.
As expected, single photonic wires show much better values than determined from
ensemble experiments. Constructs with three and four fluorophores yielded up to
~90% of transfer efficiencies, and ~70% were observed for five fluorophores.

Future Experimental Research on Photonic Wires

Single-molecule techniques have proven their applicability to aid investigation of the
working principle of complex energy transfer constructs as photonic wires. Important
characteristics of such complexes were unravelled, e.g. interaction of fluorophores
with the scaffold DNA, inefficient energy transfer steps, leakages along the energy
transfer chain and unfavourable hybridisation deficiencies.

An improved understanding of these molecular complexes can be obtained by
extending the present set-up at several points. First, alternating-laser excitation
(ALEX) offers the possibility to access structural and interaction information at the
same time [Kapanidis et al., 2004 and 2005]. Experiments on single-molecule FRET
constructs allowed further discrimination of FRET-species by introducing a
stoichiometry factor. In the case of photonic wires, fast alternation between two (or
possibly more) excitation laser wavelengths on a µs timescale could probe the state
of different fluorophores along the energy transfer chain. This would present an
easier approach to identify leakages on the one hand, but also to determine stepwise
energy transfer efficiencies on the other hand. If multiple laser excitation is
envisaged, a well-balanced combination of dichroic beamsplitters and bandpass
filters must be used.

                                 Conclusion and Outlook

A slightly different approach for studies on photonic wires includes the use of
spectrographs and CCD-cameras. With the high fluorescence intensity observed for
some constructs, it may be envisaged to collect spectra of photonic wires. This
represents a well-suited method to obtain information about the contributions from
individual fluorophores along the chain. A high quantum yield is required for suitable
CCD devices, and a sufficient number of photons must be emitted by the molecular

From the synthetic point of view, DNA represents an ideal scaffold to design optical
devices as multistep energy transfer constructs. In combination with post-labelling
strategies, advanced purification methods and hybridisation, well-defined geometries
can be realized. To work unidirectional, an energy cascade from high energy at the
input unit to lower energies along the chain of fluorophores was designed. Overall, a
spatial range of 13.6 nm over a wavelength range of ~250 nm was realized. To
extend both parameters and still work with optimal distances between fluorophores
for FRET, more fluorescent dyes could be introduced. Since the spectral bandwidth
for fluorophores is limited to avoid possible two-photon or S0-S2 excitations, an
increasing number of fluorophores results in shorter distance between absorption
and emission spectra. As a result, the analysis of energy transfer steps gets more
difficult and ambiguous. One possible solution at this point could be the casual
application of homo-FRET steps along the chain, i.e. energy transfer between two
identical fluorophores.

Finally, a desirable extension of photonic wires would be a switching element for
energy transfer. This could be realized by a combination with a molecule which can
be addressed by chemical or electrical influences or light. Hereby, the molecule
undergoes a reversible transition from a non-absorbing state to an absorbing state
with negligible fluorescence quantum yield. Concerning the mechanism for switching,
chemical methods would require a flow chamber and suffer from slow reaction times.
Light-induced switching requires photoinduced reactions (e.g. cyclization) which can
be excited exclusively, i.e. without exciting any chromophoric unit along a photonic
wire. The generation of an electric field usually affects larger volumes and does not
allow exclusive switching of one single photonic wire.

                                 Conclusion and Outlook

5.2. Single-Molecule Photoswitch: A Mechanistic View

The first experiments in preparing a reversible dark-state of a single fluorescent
molecule were carried out in 1992 [Basché and Moerner, 1992]. At the cryogenic
temperature of liquid helium, they demonstrated the process of “spectral hole-
burning” at a single perylene molecule embedded in a host crystal of polyethylene.

At room temperature, the first molecule which exhibited reversible activation of a dark
state was derived directly from nature, i.e. the chromophoric unit of the green
fluorescent protein (GFP) of aequorea victoria [Dickson et al., 1997]. The
chromophoric unit is spontaneously formed by three amino acids, i.e. glycine,
tyrosine and threonine (or serine), totally shielded from the surrounding environment
by a cylindrical barrel of the protein backbone. Responsible for the observation of
fluorescence switching is a reversible proton exchange at the chromophoric unit of
the protein.

Many approaches of switching single molecules were done by exploiting
photoinduced chemical reactions, e.g. cyclization-reactions of an arrangement of π-
electrons forming or breaking bonds [Chibisov and Görner, 1997 and 1997b]. At the
single-molecule level, Irie and coworkers used such a light-induced cyclization
reaction to create an internal quenching unit which absorbs fluorescence light via
FRET and hereby “turns” fluorescence off [Irie et al., 2002]. The reaction is
reversible, and the donor molecule exhibits fluorescence again after a few seconds.

This work represents the first successful single-molecule photoswitch at room
temperature using a commercially-available fluorophore [Heilemann et al., 2005].
Fluorescence of carbocyanine derivatives was reproducibly restored by applying two
distinct laser wavelengths. First, laser light excitation at 633 nm probed fluorescence
and prepared a dark-state. In a second step, excitation at 488 nm restored the
fluorescent state. A mechanistic description of this process was given in the
experimental part of this work (see 4.4). The reversible dark-state is only observed in
the absence of oxygen and requires the presence of a thiol-reagent at higher
concentration. This leads to the interpretation that competing reactions leading to

                                  Conclusion and Outlook

photobleaching are more likely for carbocyanines and have to be excluded
imperatively. Furthermore, the presence of thiol-containing reagents is required:
acting as triplet-quenchers for carbocyanine derivatives, they must - in a more
general description - be regarded as potential electron-donors.

Experiments with other fluorophores demonstrated the importance of the redox
properties of the environment on the formation of dark-states (see 4.1.2-4.1.5). It was
demonstrated that fluorescence intermittencies (i) can be suppressed in reductive
environment for carborhodamine derivatives (see 4.1.4) or (ii) can be generated in a
similar environment for oxazine derivatives (see 4.1.3). Yet, a direct proof of this
theory by detecting a radical anion or cation species could not be furnished. In
context with fluorescence switching of carbocyanine dyes, an intermediate radical ion
state which plays a key role in this process may be expected. This would be in
agreement with other observations made for carbocyanine dyes and very recently
published: in the presence of the short-wavelength absorbing fluorophore Cy3, the
fluorescence of the reporter fluorophore Cy5 could be turned “on” and “off”, requiring
very short distances below 3 nm [Bates et al., 2005]. It can be assumed that Cy3 acts
as an electron-donor and provides the electron to neutralize a radical cation state of
Cy5, explaining both the pre-requisite of very short distances and an exponentially-
shaped distance dependence. In agreement with the experimental conditions
determined in this work (see 4.1.2), experiments were carried out in the absence of
oxygen. Contrarily, the presence of potassium iodide was required, which is known to
enhance the triplet formation by the heavy-atom effect [Kasha, 1952]. This should be
viewed with similar caution as the effect of thiol-reagents: triplet-depopulating
properties and redox-properties cannot be separated and depend on the molecular
species under investigation. Iodine promotes triplet transitions and provides electrons
as reductive agents, together with fluorescence quenching at higher concentration.
As a consequence, the role of the triplet state may be smaller than assumed from the
first results, and a radical state is more probable. Other work published recently
shows that the fluorescent state of the structurally closely related carbocyanine
Alexa647 [Buschmann et al., 2003] could be manipulated by generation of an electric
field [White et al., 2004]. Together with a donor fluorophore, the carbocyanine dye
was attached to 40bp double-stranded DNA in direct proximity and at a distance of
40 base pairs, i.e. out of range for the observation of FRET. A solution of this labelled
DNA was measured at the output of a nanopipette with two-colour excitation and

                                 Conclusion and Outlook

controllable electric field. The resulting FRET-efficiency showed a pronounced
dependency on the magnitude of the electric field, and a nonfluorescent state of
Alexa647 was observed at a negative potential below –0.2 V. The process was only
observed at the high electric field generated at the output tip of the nanopipette and
was highly reversible.

Published data from three different groups with a similar arrangement of fluorophores
and reversible switching of the fluorescent state are summarized in figure 5-2: part
(A) shows a carbocyanine dye under two-laser excitation and typical experimental
conditions as presented in this work. Molecules switching “on” from a dark state are
observed, as well as fluorescence trajectories exhibit fluctuations between dark
states and the fluorescent states. In (B), the approach by Zhuang and coworkers is
depicted, requiring a neighbouring fluorophore and exhibiting a short-scaled distance
dependence, which exhibits similarities to electron transfer processes. In (C),
switching of fluorescence by the generation of an electric field is demonstrated, as
published by Klenerman and coworkers. The most evident similarity of all three
approaches is the need for an electron donor or an electric field. Photooxidation by
oxygen and triplet state formation are competing processes and can be neglected.

Nevertheless, the question of the molecular mechanism causing fluorescence
switching in carbocyanines remains. The dark-state created upon irradiation at 633
nm was probed for its absorption properties using smFRET. It could be demonstrated
that more than one intermediate state is involved, one in resonance with a donor
emission, and a second essentially nonabsorbing one. Further, fluorescence “off”-
states of Cy5 with increased efficiency of donor quenching compared to FRET were
observed. As a consequence, one can assume the presence of different species
which exhibit structural changes in the basic chromophoric unit, i.e. in the number of
conjugated π-electrons.

Experiments carried out at the ensemble level did not show any significant
contribution to the absorption spectrum after preparing the dark state. Reflecting that
extinction coefficients for most organic compounds are quite low (as they represent
the probability of an absorption process) compared to a value of ~2.5⋅105 M-1cm-1 for
the maximum absorption wavelength of Cy5, the absorption of the dark states
involved may be below the detection limit of absorption spectroscopy and cannot be
discriminated from background absorption.

                                 Conclusion and Outlook

                                                             Figure 5-2: Three recently
                                                             published        methods             for
                                                             reversible       switching           of
                                                             fluorescence                       from
                                                             carbocyanine           dyes.        (A)
                                                             Fluorescence switching of
                                                             Cy5 presented in this work
                                                             (see     4.4),        (B)        optical
                                                             switching of Cy5 in the
                                                             proximity        of     a        donor
                                                             fluorophore                  exhibits
                                                             shorter                     distance
                                                             dependency            than        FRET
                                                             and was used as short-
                                                             range spectroscopic ruler,
                                                             (C) an electric field at the
                                                             tip of a nanopipette is used
                                                             to manipulate the FRET
                                                             efficiency        between             a
                                                             nearby           donor              and
                                                             carbocyanine                acceptor
                                                             molecule     (red:          acceptor
                                                             emission,        green:          donor
                                                             emission),       generating           a
                                                             dark      state             of      the
                                                             carbocyanine at negative

Since thermal stability of the dark states which are involved in fluorescence switching
is observed for at least a few hours at room temperature, other methods to
characterize newly formed species might be used. The most difficult barrier
imaginable for most methods is the low concentration. Typical working conditions are
solutions of 10-6 M of the dye in PBS (~10-3 mg/ml), which excludes structural
methods such as nuclear magnetic resonance (NMR) spectroscopy or UV/VIS/IR-
spectroscopy. Mass spectroscopic methods are critical because of possibly
“extreme” conditions at the injection step and low concentrations, too. Although
exhibiting high-pressure or current, HPLC or capillary electrophoresis (CE) could be
used for separating compounds, but lacks a detection method that could be applied
for dark state intermediates. Single-paired electrons can be observed by electron-
                                  Conclusion and Outlook

spin resonance, but again concentrations of about 10- to 100fold higher are required.
In summary, the fact of low concentration and short-time thermal stability together
with poor detection possibilities makes a further characterization very difficult.

It seems more promising to think about single-molecule techniques which could
probe intermediate dark states. Different redox-properties as well as radical-
stabilizing reagents could be envisaged. Single-molecule FRET constructs with
different distances, excitation sources and excitation modulation or polarization
detection could reveal further information.

Experimental conditions chosen in this work for switching experiments on
carbocyanines required oxygen-free environment and the presence of a thiol-moiety
as reducing agent. These conditions are frequently used in smFRET-assays for
biological applications, since the increased stability of the fluorophores dramatically
extends the observation time [Ha et al., 2002]. It is interesting to note that the
observation of reversible dark state formation of carbocyanine dyes was reported
previously in such smFRET experiments [Blanchard et al., 2004], and were attributed
to photophysical phenomena. It is therefore intuitively easy to understand that
smFRET-experiments have to be interpreted carefully if carbocyanine dyes are
involved. Besides a large spectrum of photophysical reactions, the observation of
reversible photoactivation of a fluorescent dark state for carbocyanine dyes
represents a further complication which must be taken into account.

Besides further investigation of the switching mechanism observed for carbocyanine
dyes, one might think of potentially interesting applications of this method. Optical
data storage can, of cause, be envisaged, supported by the observed reversibility
and high rate of reproducibility. If one molecule is used to store one bit, data storage
is only limited by the addressability of this molecule: if a focussed laser is used, the
shape of the excitation profile limits the surface density to ~4 molecule/µm2 (or a
spacing of at least 0.5 µm between adjacent molecules) to avoid excitation of more
than one molecule. This would yield a storage capacity of 4x108 bits/cm2, i.e. 48
Megabyte. If a better addressability by confined excitation profiles is realized, an
even higher density of data storage can be realized. To finally get closer to such
applications, experimental conditions have to be modified. Research effort could be
focussed on suited polymer materials which do not incorporate molecular oxygen,
such as polyvinylalcohol (PVA) or others. Chemical modification of the polymer could

                                  Conclusion and Outlook

focus on elaborating a similar microenvironment for the fluorophores and make
photoswitching possible for incorporated molecules. These approaches could lead to
a convenient data storage material with easy handling.

Another application possible is the selective activation of a fluorophore in living cells.
Fluorescence recovery after photobleaching and optical highlighting of fluorescent
proteins can provide insights into the diffusive or directed movement of proteins and
track rapid protein behaviour [Lippincott-Schwartz, 2001; Ando et al., 2004]. Similar
to fusion proteins used so far, conventional fluorophores may be envisaged for this
purpose as well. The photoactivation process of carbocyanines derivatives presented
in this work can offer an alternative method for selective highlighting biomolecules in
cells which cannot be tagged with a fluorescent protein.

6. References

A. P. Alivisatos. The use of nanocrystals in biological detection. Nature Biotechnol.,
   2004, 22, 47-52.

W. P. Ambrose, P. M. Goodwin, J. H. Jett, A. Van Orden, J. H. Werner, R. A. Keller.
   Single Molecule Fluorescence Spectroscopy at Ambient Temperature. Chem.
   Rev., 1999, 99, 2929-2956.

R. Ando, H. Mizuno, A. Miyawaki. Regulated Fast Nucleocytoplasmic Shuttling
   Observed by Reversible Protein Highlighting. Science, 2004, 306, 1370-1373.

J. Anzai, T. Hoshi, T. Osa. Controlled deposition of glucose oxidase on platinum
   electrode based on a avidin/biotin system for the regulation of output current of
   glucose sensors. Anal. Chem., 1995, 15, 770-774.

P. W. Atkins. Physikalische Chemie. VCH Weinheim, 1995.

R. M. Badger, A. C. Wright, R. F. Whitlock. Absolute Intensities of the Discrete and
   Continuous Absorption Bands of Oxygen Gas at 1.26 and 1.065 µm and the
   Radiative Lifetime of the 1∆g state of Oxygen. J. Chem. Phys., 1965, 43, 4345-

G. S. Baird, D. A. Zacharias, R. Y. Tsien. Biochemistry, mutagenesis, and
   oligomerisation of DsRed, a red fluorescent protein from coral. PNAS, 2000, 97,

T. Basché, W. E. Moerner. Optical modification of a single impurity molecule in a
   solid. Nature, 1992, 355, 335-337.

M. Bates, T. R. Blosser, X. Zhuang. Short-range spectroscopic ruler based on single-
   molecule optical switch. Phys. Rev. Lett., 2005, 94, 10801-10804.

J. Bernard, L. Fleury, H. Talon, M. Orrit. Photon Bunching in the Fluorescence From
   Single Molecules – a Probe for Intersystem Crossing. J. Chem. Phys., 1993, 98,

R. A. Bissell, E. Cordova, A. E. Kaiser, J. F. Stoddart. A chemically and
   electrochemically switchable molecular shuttle. Nature, 1994, 369, 133-137.


S. C. Blanchard, R. L. Gonzalez Jr., H. D. Kim, S. Chu, J. D. Puglisi. tRNA selection
   and kinetic proofreading in translation. Nat. Struct. Mol. Biol., 2004, 11, 1008-

M. Boehmer, J. Enderlein. Fluorescence spectroscopy of single molecules under
   ambient conditions: methodology and technology. ChemPhysChem, 2003, 4, 793-

M. A. Bopp, Y. Jia, L. Li, R. J. Cogdell, R. M. Hochstrasser. Fluorescence and
   photobleaching dynamics of single light-harvesting complexes. PNAS, 1997, 94,

M. Born, E. Wolf. Principles of Optics, 7th edition. Cambridge University Press, 2002.

V. Buschmann, K. D. Weston, M. Sauer. Spectroscopic Study and Evaluation of Red-
   Absorbing Fluorescent Dyes. Bioconjugate Chem., 2003, 14, 195-204.

C. Bustamante, J. C. Macosko, G. J. Wuite. Grabbing the cat by the tail: manipulating
   molecules one by one. Nat. Rev. Mol. Cell. Biol., 2000, 1, 130-136.

F. K. Chan, R. M. Siegel, D. Zacharias, R. Swofford, K. L. Holmes, R. Y. Tsien, M. J.
   Lenardo. Fluorescence resonance energy transfer analysis of cell surface
   receptor interactions and signalling using spectral variants of the green
   fluorescent protein. Cytometry, 2001, 44, 361-368.

Y. Chen, A. Periasamy. Characterization of two-photon excitation fluorescence
   lifetime imaging microscopy for protein localization. Microscopy Research and
   Technique, 2004, 63, 72-80.

A. K. Chibisov, H. Görner. Photoprocesses in Spiropyran-Derived Merocyanines. J.
   Phys. Chem. A, 1997, 101, 4305-4312.

A. K. Chibisov, H. Görner. Singlet versus triplet photoprocesses in
   indodicarbocyanine dyes and spiropyran-derived merocyanines. J. Photochem.
   Photobiol., 1997b, 105 261-267.

G. Chirico, F. Cannone, A. Diaspro, S. Bologna, V. Pellegrini, R. Nifosi, F. Beltram.
   Multiphoton switching dynamics of single green fluorescent proteins. Phys. Rev.
   E, 2004, 70, 030901.

T. Christ, F. Kulzer, P. Bordat, T. Basché. Watching the Photooxidation of a Single
   Molecule. Ang. Chem. Int. Ed., 2001, 113, 4323-4326.


D. M. Chudakov, V. V. Verkhusha, D. B. Staroverov, E. A. Souslova, S. Lukyanov, K.
   A. Lukyanov. Photoswitchable cyan fluorescent protein for protein tracking.
   Nature Biotechnology, 2004, 11, 1435-1439.

J.-P. Clamme, A. A. Deniz. Three-Color Single-Molecule Fluorescence Resonance
   Energy Transfer. ChemPhysChem, 2004, 5, 1-4.

R. M. Clegg. Fluorescence resonance energy transfer and nucleic acids. Methods in
   Enzymology, 1992, 211, 353-388.

R. M. Clegg, A. I. H. Murchie, A. Zechel, C. Carlberg, S. Diekmann, D. M. J. Lilley.
   Fluorescence Resonance Energy Transfer Analysis of the Structure of the Four-
   Way DNA Junction. Biochemistry, 1992b, 31, 4846-4856.

R. M. Clegg. Fluorescence resonance energy transfer. Curr. Opin. Biotechnol., 1995,
   6, 103-110.

E. U. Condon. Nuclear motions associated with electron transitions in diatomic
   molecules. Physical Review, 1928, 32, 858-872.

R. E. Dale, J. Eisinger, W. E. Blumberg. The orientational freedom of molecular
   probes. The orientation factor in intramolecular energy transfer. Biophysical
   Journal, 1979, 26, 161-194.

A. A. Deniz, M. Dahan, J. R. Grunwell, T. Ha, A. E. Faulhaber, D. S. Chemla, S.
   Weiss, and P. G. Schultz. Single-pair fluorescence resonance energy transfer on
   freely diffusing molecules: observation of Förster distance dependence and
   subpopulations. PNAS, 1999, 96, 3670-3675.

D. L. Dexter. A theory of sensitised luminescence in solids. J. Chem. Phys., 1953,
   21, 836.

J. Deisenhofer, O. Epp, K. Miki, R. Huber, H. Michel. X-ray structure analysis of a
   membrane protein complex. Electron density map at 3 Å resolution and a model
   of the chromophores of the photosynthetic reaction center from
   Rhodopseudomonas viridis. J. Mol. Biol., 1984, 180, 385-398.

R. M. Dickson, A. B. Cubitt, R. Y. Tsien, W. E. Moerner. On/off blinking and switching
   behaviour of single molecules of green fluorescent protein. Nature, 1997, 388,


A. Dietrich, V. Buschmann, C. Müller, M. Sauer. Fluorescence Resonance Energy
   Transfer (FRET) and competing processes in donor-acceptor substituted DNA
   strands: a comparative study of ensemble and single-molecule data. Rev. Mol.
   Biotechn. 2002, 82, 211-231.

M. Diez, B. Zimmermann, M. Börsch, M. König, E. Schweinberger, S. Steigmiller, R.
   Reuter, S. Felekyan, V. Kudryavtsev, C. A. M. Seidel, P. Gräber. Proton-powered
   subunit rotation in single membrane-bound F0F1-ATP synthase. Nature structural
   & molecular biology, 2004, 2, 135-141.

Z. J. Donhauser, B. A. Mantooth, K. F. Kelly, L. A. Bumm, J. D. Monnell, J. J.
   Stapleton, D.W. Price, A. M. Rawlett, D. L. Allara, J. M. Tour, P. S. Weiss.
   Conductance switching in single molecules through conformational changes.
   Science, 2001, 292, 2303-2307.

N. Dontha, W. B. Nowall, W. G. Kuhr. Generation of Biotin/Avidin/Enzyme
   Nanostructures with Maskless Photolithography. Anal. Chem., 1997, 69, 2619-

S. Doose, H. Neuweiler, M. Sauer. A close look at fluorescence quenching of organic
   dyes by tryptophan. ChemPhysChem, 2005, accepted.

K. H. Drexhage. Structure and Properties of Laser Dyes. Topics in Applied Physics –
   Dye Lasers – Springer Verlag Berlin-Heidelberg-New York, 1973, 144-179.

M. Eigen. R. Rigler. Sorting Single Molecules: applications to diagnostics and
   evolutionary biotechnology. PNAS, 1994, 91, 5740-5747.

C. Eggeling, J. Widengren, R. Rigler, C. A. M. Seidel. Photobleaching of Fluorescent
   Dyes under Conditions Used for Single-Molecule Detection: Evidence of Two-
   Step Photolysis. Anal. Chem., 1998, 70, 2651-2659.

C. Eggeling, J. R. Fries, L. Brand, R. Günther, C. A. M. Seidel. Monitoring
   conformational dynamics of a single molecule by selective fluorescence
   spectroscopy. PNAS, 1998b, 95, 1556-1561.

C. Eggeling, A. Volkmer, C. A. M. Seidel. Photobleaching Kinetics of Rhodamine 6G
   by One- and Two-Photon Induced Confocal Fluorescence Microscopy.
   ChemPhysChem, 2005, 6, 791-804.


S. A. Empedocles, R. Neuhauser, M. G. Bawendi. Three-dimensional orientation
   measurements of symmetric single chromophores using polarization microscopy.
   Nature, 1999, 399, 126-130.

J. Enderlein, P. M. Goodwin, A. Van Orden, W. P. Ambrose, R. Erdmann, R. A.
   Keller. A maximum likelihood estimator to distinguish single molecules by their
   fluorescence decays. Chem. Phys. Letters, 1997, 270, 464-470.

J. Enderlein, I. Gregor, D. Patra, J. Fitter. Art and Artefacts of Fluorescence
   Correlation Spectroscopy. Curr. Pharm. Biotechnol., 2004, 5, 155-161.

D. S. English, E. J. Harbron, P. F. Barbara. Probing Photoinduced Intersystem
   Crossing by Two-Color, Double Resonance Single Molecule Spectroscopy. J.
   Phys. Chem. A, 2000, 104, 9057-9061.

T. Förster. Zwischenmolekulare Energiewanderung und Fluoreszenz. Annalen der
   Physik, 1948, 2, 55-75.

T. Fukaminato, T. Sasaki, N. Tamai, M. Irie. Digital photoswitching of fluorescence
   based on the photochromism of diarylethene derivatives at a single-molecule
   level. J. Am. Chem. Soc., 2004, 126, 14843-14849.

T. Funatsu, Y. Harada, M. Tokunaga, K. Saito, T. Yanagida. Imaging of single
   fluorescent molecules and individual ATP turnovers by single myosin molecules in
   aqueous solution. Nature, 1995, 374, 555-559.

E. Füreder-Kitzmüller, J. Hesse, A. Ebner, H. J. Gruber, G. J. Schütz. Non-
   exponential bleaching of single bioconjugated Cy5 molecules. Chem. Phys. Lett.,
   2005, 404, 13-18.

E. Galperin, V. V. Verkhusha, A. Sorkin. Three-chromophore FRET microscopy to
   analyze multiprotein interactions in living cells. Nature Methods, 2004, 1, 209-217.

L. Giordano, T. M. Jovin, M. Irie, E. A. Jares-Erijman. Diheteroarylethenes as
   thermally stable photoswitchable acceptors in photochromic fluorescence
   resonance energy transfer (pcFRET). J. Am. Chem. Soc., 2002, 124, 7481-7489.

D. I. Gittins, D. Bethell, D. J. Schiffrin, R. J. Nichols. A nanometre-scale electronic
   switch consisting of a metal cluster and redox-addressable groups. Nature, 2000,
   408, 67-69.


A. N. Glazer. Light guides. Directional energy transfer in a photosynthetic antenna. J.
   Biol. Chem., 1989, 264, 1-4.

P. M. Goodwin, W. P. Ambrose, R. A. Keller. Single-Molecule Detection in Liquids by
   Laser-Induced Fluorescence. Acc. Chem. Res., 1996, 29, 607-613.

T. Ha, T. Enderle, D. F. Ogletree, D. S. Chela, P. R. Selvin, S. Weiss. Probing the
   Interaction between two Single Molecules: Fluorescence Resonance Energy
   Transfer between a single donor and a single acceptor. PNAS, 1996, 93, 6264-

T. Ha, A. Y. Ting, J. Liang, A. A. Deniz, D. S. Chela, P. G. Schultz, S. Weiss.
   Temporal fluctuations of fluorescence resonance energy transfer between two
   dyes conjugated to a single protein. Chem. Phys., 1999, 247, 107-118.

T. Ha. Single-Molecule fluorescence resonance energy transfer. Methods, 2001, 25,

T. Ha, I. Rasnik, W. Cheng, H. P. Babcock, G. H. Gauss, T. M. Lohman, S. Chu.
   Initiation and re-initiation of DNA unwinding by the Escherichia coli Rep helicase.
   Nature, 2002, 419, 638-641.

T. Ha and J. Xu. Photodestruction Intermediates Probed by an Adjacent Reporter
   Molecule. Phys. Rev. Lett., 2003, 90, 22320021-22320024.

K.-T. Han, private communication, 2005.

R. Harnbury-Brown, R. Q. Twiss. Nature, 1956, 177, 27.

R. P. Haugland. Optical Microscopy for Biology, Wiley-Liss, New-York, 1989, 147-

E. Haustein, P. Schwille. Ultrasensitive Investigations of biological systems by
   fluorescence correlation spectroscopy. Methods, 2003, 29, 153-166.

M. Heilemann, D. P. Herten, R. Heintzmann, C. Cremer, C. Müller, P. Tinnefeld, K. D.
   Weston, J. Wolfrum, M. Sauer. High-resolution colocalization of single dye
   molecules bye fluorescence lifetime imaging microscopy. Analytical Chemistry,
   2002, 74, 3511-3517.

M. Heilemann, P. Tinnefeld, G Sanchez-Mosteiro, M. Garcia-Parajó, N. F. van Hulst.
   Multistep Energy Transfer in Single Molecular Photonic Wires. J. Am. Chem.
   Soc., 2004, 126, 6514-6515.

M. Heilemann, E. Margeat, R. Kasper, M. Sauer, P. Tinnefeld. Carbocyanine Dyes as
   Efficient Reversible Single-Molecule Optical Switch. J. Am. Chem. Soc., 2005,
   127, 3801-3806.

D. P. Herten, P. Tinnefeld, M. Sauer. Identification of singly fluorescently labelled
   mononucleotide molecules in solution by spectrally resolved time-correlated
   single photon counting. Appl. Physics B, 2000, 71, 765-771.

T. Heinlein, J.-P. Knemeyer, O. Piestert, M. Sauer. Photoinduced Electron Transfer
   between Fluorescent Dyes and Guanosine Residues in DNA-Hairpins. J. Phys.
   Chem. B., 2003, 107, 7957-7964.

T. Heinlein, A. Biebricher, P. Schlüter, C. Roth, D. P. Herten, J. Wolfrum, M.
   Heilemann, C. Müller, P. Tinnefeld, M. Sauer. Nanometer-localized fluorescence
   imaging and counting of single molecules with spectrally-resolved fluorescence
   lifetime imaging microscopy (SFLIM). ChemPhysChem, 2005, 6, 949-955.

T. Hirschfeld. Optical microscopic observation of single small molecules. Appl. Optics
   1976, 15, 2965.

J. Hofkens, M. Maus, T. Gensch, T. Vosch, M. Cotlet, F. Kohn, A. Herrmann, K.
   Müllen, F. de Schryver. Probing photophysical processes in individual
   multichromophoric dendrimers by single-molecule spectroscopy. J. Am. Chem.
   Soc., 2000, 122, 9278-9288.

J. Hofkens, M. Cotlet, T. Vosch, P. Tinnefeld, K. D. Weston, C. Ego, A. Grimsdale, K.
   Muellen, D. Beljonne, J. L. Bredas, S. Jordens, G. Schweitzer, M. Sauer, F. C. De
   Schryver. Revealing competitive Förster-type resonance energy transfer
   pathways in single bichromophoric molecules. PNAS, 2003, 100, 13146-13151.

S. Hohng, C. Joo, T. Ha. Single-molecule three-color FRET. Biophysical Journal,
   2004, 87, 1328-1337.

S. Hohng, T. Ha. Near-Complete Suppression of Quantum Dot Blinking in Ambient
   Conditions. J. Am. Chem. Soc., 2004, 126, 1324-1325.

S. Hohng, T. Ha. Single-Molecule Quantum Dot Fluorescence Resonance Energy
   Transfer. ChemPhysChem, 2005, 6, 1-5.

X. Hu, K. Schulten. Model for the light-harvesting complex I (B875) of Rhodobacter
   sphaeroides. Phys. Today, 1997, 50, 28-34.


Z. Huang, D. Ji, A. Xia, F. Koberling, M. Patting, R. Erdmann. Direct Observation of
   Delayed Fluorescence from a Remarkable Back-Isomerization in Cy5. J. Am.
   Chem. Soc., 2005, ASAP Articale.

T. Huser, M. Yan, L. J. Rothberg. Single chain spectroscopy of conformational
   dependence of conjugated polymer photophysics. PNAS, 2000, 97, 11187-11191.

M. Irie, T. Fukaminato, T. Sasaki, N. Tamai, T. Kawai. A digital fluorescent molecular
   photoswitch. Nature, 2002, 420, 759-760.

E. A. Jares-Erijman, T. M. Jovin. FRET Imaging. Nat. Biotechnol., 2003, 21, 1387-

G. Jung, J. Wiehler, A. Zumbusch. The photophysics of green fluorescent protein:
   influence of the key amino acids at positions 65, 204, and 222. Biophys. J., 2005,
   88, 1932-1947.

A. N. Kapanidis, N. K. Lee, T. A. Laurence, S. Doose, E. Margeat, S. Weiss.
   Fluorescence-aided molecule sorting : Analysis of structure and interactions by
   alternating-laser excitation of single molecules. PNAS, 2004, 101, 8936-8941.

A. N. Kapanidis, T. A. Laurence, N. K. Lee, E. Margeat, X. Kong, S. Weiss.
   Alternating-Laser Excitation of Single Molecules. Acc. Chem. Res., 2005, ASAP

M. Kasha. J. Phys. Chem., 1952, 20, 71.

B. S. T. Kasibhatla, A. P. Labonte, F. Zahid, R. G. Reifenberger, S. Datta, C. P.
   Kubiak. Reversibly altering electronic conduction through a single molecule by a
   chemical binding event. J. Phys. Chem. B, 2003, 107, 12378-12382.

J. T. M. Kennis, D. S. Larsen, I. H. M. van Stokkum, M. Vengris, J. J. van Thor, R.
   van Grondelle. Uncovering the hidden ground state of green fluorescent protein.
   PNAS, 2004, 101, 17988-17993.

C. Kluge, T. Seidel, S. Bolte, S. S. Sharma, M. Hanitzsch, B. Hanitzsch. B. Satiat-
   Jeunemaitra, J. Ross, M. Sauer, D. Golldack, K. J. Dietz.. Subcellular distribution
   of the V-ATPase complex in plant cells, and in vivo localisation of the 100 kDa
   subunit VHA-a within the complex. BMC Cell Biol., 2004, 5, 1-18.

J. P. Knemeyer, N. Marmé, M. Sauer. Probes for Detection of Specific DNA
   Sequences at the Single-Molecule Level. Anal. Chem., 2000, 72, 3717-3724.


F. Köhn, J. Hofkens, U. M. Wiesler, M. Cotlet, M. van der Auweraer, K. Müllen, F. C.
   DeSchryver. Single-molecule spectroscopy of a dendrimer-based host-guest
   system. Chemistry, 2001, 7, 4126-4133.

H. Kuhn. Progress in the chemistry of Organic Natural Products, Vol. 16. Springer
   Verlag, Wien, 1959, 411ff.

T. D. Lacoste, X. Michalet, F. Pinaud, D. S. Chemla, A. P. Alivisatos, S. Weiss.
   Ultrahigh-resolution multicolor colocalization of single fluorescent probes. PNAS,
   2000, 97, 9461-9466.

J. R. Lakowicz, Principles of Fluorescence Spectroscopy, Second Edition. Plenum
   Press, New York, 1999.

V. LeTilly, C. A. Royer. Fluorescence anisotropy assays implicate protein-protein
   interactions regulating trp repressor DNA binding. Biochemistry, 1993, 32, 7753-

Y. C. Liang, A. S. Dvornikov, P. M. Rentzepis. Nonvolatile read-out molecular
   memory. PNAS, 2003, 100, 8109-8112.

J. Lippincott-Schwartz, E. Snapp, A. Kenworthy. Studying protein dynamics in living
   cells. Nature Rev. Mol. Cell Biol., 2001, 6, 444-456.

J. Liu, Y. Lu. FRET study of a trifluorophore-labeled DNAzyme. JACS, 2002, 124,

J. J. Macklin, J. K. Trautman, T. D. Harris, L. E. Brus. Imaging and Time-Resolved
   Spectroscopy of Single Molecules at an Interface. Science, 1996, 272, 255-258.

N. Marmé, J. P. Knemeyer, M. Sauer, J. Wolfrum. Inter- and Intramolecular
   Fluorescence Quenching of Organic Dyes by Tryptophan. Bioconj. Chem., 2003,
   14, 1133-1139.

T. B. McAnaney, W. Zeng, C. F. E. Doe, N. Bhanji, S. Wakelin, D. S. Pearson, P.
   Abbyad, X. Shi, S. G. Boxer, C. R. Bagshaw. Protonation, photobleaching and
   photoactivation of yellow fluorescent protein (YFP 10C): a unifying mechanism.
   Biochemistry, 2005, ASAP article.

P. B. Merkel, D. R. Kearns. Remarkable Solvents effects on the lifetime of 1∆g
   oxygen. J. Am. Chem. Soc., 1972, 1029-1030.


U. Mets, R. Rigler. Submillisecond Detection of Single Rhodamine Molecules in
   Water. J. Fluoresc., 1994, 4, 259-264.

X. Michalet, T. D. Lacoste, S. Weiss. Ultrahigh-resolution colocalization of spectrally
   separable point-like fluorescent probes. Methods, 2001, 25, 87-102.

X. Michalet, F. F. Pinaud, L. A. Bentolila, J. M. Tsay, S. Doose, J. J. Li, G.
   Sundaresan, A. M. Wu, S. S Gambhir, S. Weiss. Quantum dots for live cells, in
   vivo imaging and diagnostics. Science, 2005, 307, 538-544.

M. Minsky. Microscopy apparatus. US patent 3013467, filed 1957.

W. E. Moerner, L. Kador. Optical Detection and spectroscopy of single molecules in a
   solid. Phys. Rev. Letters, 1989, 62, 2535-2538.

W. E. Moerner, M. Orrit. Illuminating Single Molecules in Condensed Matter. Science,
   1999, 283, 1670-1676.

W. E. Moerner, D. P. Fromm. Methods of single-molecule fluorescence spectroscopy
   and microscopy. Rev. Sci. Instrum., 2003, 74, 3597-3619.

F. Moresco, G. Meyer, K. H. Rieder, H. Tang, A. Gourdon, C. Joachim. Recording
   intramolecular mechanics during the manipulation of a large molecule. Phys. Rev.
   Lett, 2001, 86, 672-675.

H. Murakoshi, R. Iino, T. Kobayashi, T. Fujiwara, C. Ohshima, A. Yoshimura, A.
   Kusumi. Single-molecule imaging analysis of Ras activation in living cells. PNAS,
   2004, 101, 7317-7322.

H. Neuweiler, A. Schulz, A. Vaiana, J. Smith, S. Kaul, J. Wolfrum, M. Sauer. Peptide-
   based molecular beacons for the detection of p53-autoantibodies in human sera.
   Angew. Chem. Int. Ed., 2002, 114, 4769-4773.

H. Neuweiler, M. Sauer. Using Photoinduced Charge Transfer Reactions to Study
   Conformational Dynamics of Biopolymers at the Single-Molecule Level. Curr.
   Pharm. Biotechnology, 2004, 5, 285-298.

H. Neuweiler and M. Sauer. Exploring life by single-molecule fluorescence
   spectroscopy: molecular characteristics hidden by ensemble experiments can be
   revealed by fluorescence. Anal. Chem., 2005, 77, 179A-185A.

S. Nie, R. N. Zare. Optical Detection of Single Molecules. Annu. Rev. Biophys.
   Biomol. Struct., 1997, 26, 567-596.

Y. Ohya, K. Yabuki, M. Hashimoto, A. Nakajima, T. Ouchi. Multistep fluorescence
   resonance energy transfer in sequential chromophore array constructed on oligo-
   DNA assemblies. Bioconjugate Chemistry, 2003, 14, 1057-1066.

M. Orrit, J. Bernard. Single pentacene molecules detected by fluorescence excitation
   in a p-terphenyl crystal. Phys. Rev. Lett. 1990, 65, 2716-2719.

J. Pawley. Handbook of Biological Confocal Microscopy, 2nd ed. Plenum Press, New
   York, 1995.

E. J. G. Peterman, S. Brasselet, W. E. Moerner. The Fluorescence Dynamics of
   Single Molecules of Green Fluorescent Protein. J. Phys. Chem. A, 1999, 103,

O. Piestert, H. Barsch, V. Buschmann, T. Heinlein, J.-P. Knemeyer, K. D. Weston, M.
   Sauer. A Single-Molecule Sensitive DANN Hairpin System Based on
   Intramolecular Electron Transfer. Nano. Lett., 2003, 3, 979-982.

J. Psencik, Y.-Z. Ma, J. B. Arellano, J. Hala, T. Gillbro. Excitation energy transfer
   dynamics and excited-state structure in chlorosomes of Chlorobium
   phaeobacteroides. Biophys. J., 2003, 84, 1161-1179.

J. B. Randolph and A. S. Waggoner. Stability, specificity and fluorescence brightness
   of multiply-labeled fluorescent DNA probes. Nucl. Acid Res., 1997, 25, 2923-

I. Rasnik, S. A. McKinney, T. Ha. Surfaces and Orientations: Much to FRET about?
   Acc. Chem. Res., 2005, ASAP article.

D. Rehm, A. Weller. Kinetik und Mechanismus der Elektronenübertragung bei der
   Fluoreszenzlöschung in Acetonitril. Ber. Bunsenges. Physik. Chem., 1969, 73

R. Rigler, U. Mets, J. Widengren, and P. Kask. Fluorescence correlation
   spectroscopy with high count rate and low background - analysis of translational
   diffusion. European Biophysics Journal, 1993, 22,169–175.

R. Rigler, U. Mets. Diffusion of single molecules through a gaussian laser beam.
   SPIE Vol. 21,Laser Spectroscopy of Biomolecules, 1992, 239.


M. Sauer. Entwicklung effizienter Fluoreszenzfarbstoffe für den hochempfindlichen
   Nachweis von Biomolekülen unter Einsatz zeitaufgelöster LIF-Spektroskopie.
   Dissertation Univ. Heidelberg, 1995.

M. Sauer. Single-molecule sensitive fluorescence sensors based on intramolecular
   photoinduced electron transfer. Angew. Chem. Int. Ed., 2003, 115, 1790-1793.

B. Schuler, E. A. Lipman, P. J. Steinbach, M. Kumke, W. A. Eaton. Polyproline and
   the “spectroscopic ruler“ revisited with single-molecule fluorescence. PNAS,
   2005, 102, 2754-2759.

P. R. Selvin. The renaissance of fluorescence resonance energy transfer. Nature
   Structural Biology, 2000, 7(9), 730-734.

J. Seth, V. Palaniappan, R. W. Wagner, T. E. Johnson, J. S. Lindsey, D. F. Bocian.
   Soluble Synthetic Multiporphyrin Arrays. Static Spectroscopic and
   Electrochemical Probes of Electronic Communication. J. Am. Chem. Soc., 1996,
   118, 11194-11207.

E.B. Shera, N.K. Seitzinger, L.M. Davis, R.A. Keller, S.A. Soper. Detection of single
   fluorescent molecules. Chem. Phys. Lett. 1990, 174, 553.

O. Shimomura, F. H. Johnson, Y. Saiga. Extraction, purification and properties of
   aequorin, a bioluminescent protein from the luminous hydromedusan, aequorea.
   J. Cell. Comp. Physiol., 1962, 59, 223-239.

O. Shimomura. Structure of the chromophore of Aequorea green fluorescent protein.
   FEBS Lett., 1979, 104, 220-222.

A. M. Smith, X. Gao, S. Nie. Quantum-Dot Nanocrystals for In-vivo Molecular and
   Cellular Imaging. Photochem. Photobiol., 2004, 80, 377-385.

S. Speiser. Photophysics and Mechanisms of Intramolecular Electronic Energy
   Transfer in Bichromophoric Molecular Systems: Solution and Supersonic Jet
   Studies. Chem. Rev., 1996, 96, 1953-1976.

D. Stoner-Ma, A. A. Jaye, P. Matousek, M. Towrie, S. R. Meech, P. J. Tonge.
   Observation of Excited-State Proton Transfer in Green Fluorescent Protein using
   Ultrafast Vibrational Spectroscopy. J. Am. Chem. Soc., 2005, 127, 2864-2865.

L. Stryer, R. P. Haugland. Energy transfer: a spectroscopic ruler. PNAS, 1967, 58,


P. Tinnefeld, D.-P. Herten, M. Sauer. Photophysical Dynamics of Single Molecules
   Studied by Spectrally-Resolved Fluorescence Lifetime Imaging Microscopy
   (SFLIM). J. Phys. Chem. A, 2001, 105, 7989-8003.

P. Tinnefeld, V. Buschmann, K. D. Weston, M. Sauer. Direct Observation of
   Collective Blinking and Energy Transfer in a Bichromophoric System. J. Phys.
   Chem. A, 2003, 107, 323-327.

P. Tinnefeld, J. Hofkens, D.-P. Herten, S. Masuo, T. Vosch, M. Cotlet, S. Habuchi, K.
   Müllen, F. C. De Schryver, M. Sauer. Higher-excited-state photophysical
   pathways in multichromophoric systems revealed by single-molecule
   fluorescence spectroscopy. ChemPhysChem, 2004, 5, 1786-1790.

P. Tinnefeld, M. Sauer. Branching Out of Single-Molecule Fluorescence
   Spectroscopy: Challenges for Chemistry and Influence on Biology. Angew. Chem.
   Int. Ed., 2005, 44, 2642-2671.

M. Tomschik, H. Zheng, K. van Holde, J. Zlatanova, S. H. Leuba. Fast, long-range,
   reversible conformational fluctuations in nucleosomes revealed by single-pair
   fluorescence resonance energy transfer. PNAS, 2005, 102, 3278-3283.

P. M. Torgerson, M. F. Morales. Application of the Dale-Eisinger analysis to proximity
   the contractile system. PNAS, 1984, 3723-3727.

R. Y. Tsien. The green fluorescent protein. Annu. Rev. Biochemistry, 1998, 67, 509-

N. J. Turro. Modern Molecular Photochemistry. University Science Books, Pasadena,
   USA, 1991.

D. A. Vanden Bout, W. T. Yip, D. Hu, D. K. Fu, T. M. Swager, P. F. Barbara. Discrete
   Intensity Jumps and Intramolecular Electronic Energy Transfer in the
   Spectroscopy of Single Conjugated Polymer Molecules. Science, 1997, 277,

A. M. van Oijen, M. Ketelaars, J. Köhler, T. J. Aartsma, J. Schmidt. Unraveling the
   electronic structure of individual photosynthetic pigment-protein complexes.
   Science, 1999, 285, 400-402.

A.M. van Oijen, M. Ketelaars, J. Köhler, T. J. Aartsma, J. Schmidt. Spectroscopy of
   Individuall Light-Harvesting 2 Complexes of Rhodopseudomonas acidophila:


   Diagonal Disorder, Intercomplex, Heterogeneity, Spectral Diffusion, and Energy
   Transfer in the B800 Band. Biophys. J., 2000, 78, 1570-1577.

J. A. Veerman, M. F. Garcia-Parajo, L. Kuipers, N. F. van Hulst. Time-Varying Triplet
   State Lifetimes of Single Molecules. Phys. Rev. Lett., 1999, 83, 2155-2158.

G. Viswath, M. Kasha. Confirmation of the anomalous fluorescence of azulene. J.
   Chem. Phys., 1956, 24, 574-577.

T. Vosch, M. Cotlet, J. Hofkens, K. Van Biest, M. Lor, K. D. Weston, P. Tinnefeld, M.
   Sauer, L. Latterini, K. Muellen, F. C. De Schryver. Probing Förster Type Energy
   Pathways in a First Generation Rigid Dendrimer Bearing Two Perylene Imide
   Chromophores. J. Phys. Chem. A, 2003, 107, 6920-6931.

P. Vukusic, J. R. Sambles. Photonic Structures in Biology. Nature, 2003, 424, 852-

S. Vyawahare, S. Eyal, K. Mathews, S. R. Quanke. Nanometer-scale Fluorescence
   Resonance Optical Waveguides. Nanoletters, 2004, 4, 1035-1039.

R. W. Wagner, J. S. Lindsey. A Molecular Photonic Wire. J. Am. Chem. Soc., 1994,
   116, 9759-9760.

D. Wang, E. Geva. Protein Structure and Dynamics from Single-Molecule
   Fluorescence Resonance Energy Transfer. J. Phys. Chem. B, 2005, 106, 1626-

H. M. Watrob, C. P. Pan, M. D. Barkley. Two-Step FRET as a Structural Tool. J. Am.
   Chem. Soc., 2003, 125, 7336-7343.

S. Weiss. Fluorescence spectroscopy of single biomolecules. Science, 1999, 283,

S. Weiss. Measuring conformational dynamics of biomolecules by single molecule
   fluorescence spectroscopy. Nat. Struct. Biol., 2000, 7, 724-729.

K. D. Weston, P. J. Carson, H. Metiu, S. K. Buratto. Room-temperature fluorescence
   characteristics of single dye molecules adsorbed on a glass surface. J. Chem.
   Phys., 1998, 109, 7474-7485.

S. S. White, L. Ying, S. Balasubramanian, D. Klenerman. Individual Molecules of
   Dye-Labeled DNA Act as a Reversible Two-Color Switch upon Application of an
   Electric Field. Angew. Chem. International Ed., 2004, 116, 6052-6056.

J. Widengren, R. Rigler. Mechanisms of photobleaching investigated by fluorescence
   correlation spectroscopy. Bioimaging, 1996, 4, 149-157.

J. Widengren, R. Rigler. Fluorescence correlation spectroscopy as a tool to
   investigate chemical reactions in solutions and on cell surfaces. Cell. Mol. Biol.,
   1998, 44, 857-879.

J. Widengren, P. Schwille. Characterization of Photoinduced Isomerization and Back-
   Isomerization of the Cyanine Dye Cy5 by Fluorescence Correlation Spectroscopy.
   Journal of Phys. Chem. B, 2000, 104, 6416-6432.

M. Wu, P. M. Goodwin, W. P. Ambrose, R. A. Keller. Photochemistry and
   Fluorescence Emission Dynamics of Single Molecules in Solution: B-
   Phycoerythrin. J. Phys. Chem., 1996, 100, 17406-17409.

E. Yablonovitch. Photonic crystals: semiconductors of light. Sci. Am., 2001, 285, 47-
   51 and 54-55.

P. Yan, M. W. Holman, P. Robustelli, A. Chowdhury, F. I. Ishak, D. M. Adams.
   Molecular Switch Based on a Biologically Important Redox Reaction. J. Phys.
   Chem. B, 2005, 109, 130-137.

M. Yang, D. P. Millar. Fluorescence resonance energy transfer as a probe of DNA
   structure and function. Methods Enzymol., 1997, 278, 417-444.

A. Yildiz, M. Tomishige, R. D. Vale, P. R. Selvin. Kinesin Walks Hand-Over-Hand.
   Science, 2003, 303, 676-678.

W. T. Yip, D. H. Hu, J. Yu, D. A. Vanden Bout, P. F. Barbara. Classifying the
   photophysical dynamics of single- and multiple-chromophoric molecules by single
   molecule spectroscopy. J. Phys. Chem. A, 1998, 102, 7564-7575.

J. Yu, D. H. Hu, P. F. Barbara. Unmasking electronic energy transfer of conjugated
   polymers by suppression of O2 quenching. Science, 2000, 289, 1327-1330.

M. Zhao, L. Jin, B. Chen, Y. Ding, H. Ma, D. Chen. Afterpulsing and its correction in
   fluorescence correlation spectroscopy experiments. Appl. Opt. Lett., 2003, 42,

J. Zheng, J. T. Petty, R. M. Dickson. High Quantum Yield Blue Emission from Water-
   Soluble Au8 Nanodots. JACS, 2003, 125, 7780-7781.


R. Zondervan, F. Kulzer, S. B. Orlinskii, M. Orrit. Photoblinking of Rhodamine 6G in
   Poly(vinyl alcohol): Radical Dark State Formed through the Triplet. J. Phys.
   Chem. A, 2003, 107, 6770-6776.

7. Publication List

Parts of this work were published in international scientific journals.

7.1. Publications in Scientific Journals

Heilemann, M.; Herten, D.- P.; Heintzmann, R.; Cremer, C.; Mueller, C.; Tinnefeld,
P.; Weston, K. D.; Wolfrum, J.; Sauer, M. High-resolution Colocalization of Single
Dye Molecules by Fluorescence Lifetime Imaging Microscopy. Analytical Chemistry,
2002, 74, 3511-3517.

Heilemann, M.; Buschmann, V.; Piestert, O.; Tinnefeld, P.; Weston, K. D.; Sauer, M.
Development of a molecular photonic wire by means of multiparameter single-
molecule spectroscopy. Proceedings of SPIE-The International Society for Optical
Engineering, 2003, 4962, 38-46.

Heinlein, T.; Heilemann, M.; Herten, D. P.; Mueller, C.; Tinnefeld, P.; Weston, K. D.;
Sauer, M. Spectrally-resolved fluorescence lifetime imaging microscopy (SFLIM) and
coincidence analysis: new tools to study the organization of biomolecular machines.
Proceedings of SPIE-The International Society for Optical Engineering, 2003, 4962,

Heilemann, M.; Tinnefeld, P.; Sánchez-Mosteiro, G.; García-Parajó, M. F.; van Hulst,
Niek F.; Sauer, M. Multistep Energy Transfer in Single Molecular Photonic Wires. J.
Am. Chem. Soc., 2004, 126, 6514-6515. (Highlighted by Chin, G. J. Science, 2004,
304, 1213).

Tinnefeld, P.; Buschmann, V.; Weston, K. D.; Biebricher, A.; Herten, D. P.; Piestert,
O.; Heinlein, T.; Heilemann, M.; Sauer, M. How single molecule photophysical
studies complement ensemble methods for a better understanding of chromophores
and chromophore interactions. Recent Res. Devel. Physical Chem., 2004, 7, 1-31.

Herten, D. P.; Biebricher, A.; Heilemann, M.; Heinlein, T.; Müller, C.; Schlüter, P.;
Tinnefeld, P.; Weston, K. D.; Sauer, M.; Wolfrum, J. Optical single molecule
techniques for analytical and biological applications. Recent Res. Devel. Applied
Phys. 2004, 7, 2-24.

                                      Publication List

Heilemann, M.; Margeat, E.; Kasper, R.; Sauer, M.; Tinnefeld, P. Carbocyanine
Dyes as Efficient Reversible Single-Molecule Optical Switch. J. Am. Chem. Soc.,
2005, 127, 3801-3806.

Tinnefeld, P.; Heilemann, M.; Sauer, M. Design of Molecular Photonic Wires Based
on Multistep Electronic Excitation Transfer. ChemPhysChem, 2005, 6, 217-222.

Heinlein, T.; Biebricher, A.; Schlüter, P.; Roth, C. M.; Herten, D.-P.; Wolfrum, J.;
Heilemann, M.; Müller, C.; Tinnefeld, P.; Sauer, M. High-Resolution Colocalization of
Single   Molecules   within     the   Resolution         Gap   of   Far-Field   Microscopy.
ChemPhysChem, 2005, 6, 949-955.

Sánchez-Mosteiro, G.; van Dijk, E. M. H. P.; Hernando, J.; Heilemann, M. ;
Koberling, F. ; Erdmann, R.; Sauer, M.; van Hulst, N. F.; García-Parajó, M. F.
Monitoring the extent of ET on individual DNA Based Photonic Nanowires.
manuscript in preparation.

7.2. Conference Presentations

“Design of a Molecular Photonic Wire based on Multistep Energy Transfer” (Poster)

8th International Workshop on Single Molecule Detection and Ultrasensitive Analysis
in Life Sciences (PicoQuant 2002), 22-24 September 2002, Berlin.

“Development of a molecular photonic wire by means of multiparameter single-
molecule spectroscopy” (Talk)

Photonics West 2003, 25-31 January 2003, San Jose, USA.

“Multistep Energy Transfer in Single Molecular Photonic Wires” (Talk)

10th International Workshop on Single Molecule Detection and Ultrasensitive Analysis
in Life Sciences (PicoQuant 2004), 22-24 September 2004, Berlin.

8. Abbreviations

ADC      Analogue-Digital Converter

APD      Avalanche Photodiode

BME      ß-Mercaptoethanol

CCD      Charge Coupled Device

CFD      Constant-Fraction Discriminator

DNA      Deoxyribonucleic Acid

DTT      Dithiothreitol

EE(E)T   Electron Exchange (Energy) Transfer

EOM      Electrooptical Modulator

FIFO     First-In-First-Out

FLIM     Fluorescence Lifetime Imaging Microscopy

FRET     Fluorescence Resonance Energy Transfer

FWHM     Full Width Half Maximum

GMP      Guanosine monophosphate

HOMO     Highest Occupied Molecular Orbital

HPLC     High Performance Liquid Chromatography

IRF      Instrument Response Function

LUMO     Lowest Occupied Molecular Orbital

MEA      Mercaptoethylamine

NHS      Succinimidyl Ester

PBS      Phosphate Buffered Saline

PET      Photoinduced Electron Transfer

PMT      Photomultiplier Tube

RhG      Rhodamine Green


SFLIM   Spectrally-Resolved Fluorescence Lifetime Imaging Microscopy

SMFS    Single-Molecule Fluorescence Spectroscopy

SOMO    Semi Occupied Molecular Orbital

TAC     Time-to-Amplitude Converter

TCBQ    Tetrachlorobenzoquinone

TCSPC   Time-Correlated Single-Photon Counting

TMR     Tetramethylrhodamine

TRES    Time-Resolved Emission Spectroscopy

TTTR    Time-Tagged Time-Resolved

LED     Light-Emitting Diode

9. Acknowledgements

I am grateful to all the people who helped me to realize my dissertation work in the
last three years.

I am very thankful to Prof. Markus Sauer for introducing me to the exciting field of
single-molecule fluorescence spectroscopy and the possibility to work in projects at
the cutting-edge of science.

This work would not have been possible without numerous collaborations, fruitful
discussions, and always a helping hand and a great working atmosphere. At this
point, I want to thank all the people of the “Applied Laser Physics and Laser
Spectroscopy” group in Bielefeld. Just as much, I want to thank all members of the
biophysical chemistry group in Heidelberg, I enjoyed the time I spent at the PCI
during the first half of this work.

This work would not have been possible without many helping thoughts, discussions
and inspiration by Markus Sauer, Philip Tinnefeld, Hannes Neuweiler and Sören
Doose, Kjung-Tae Han, Dirk-Peter Herten, and Christian Müller. I am thankful to Jörg
Enderlein for helping in theoretical questions. Many thanks to Johan Hofkens, Ken
Weston and Andrea Vaiana for fruitful discussions.

I am especially thankful to Gerd Wiebusch for all technical help and for an advanced
introduction into optics and lasers, to Stephan Wörmer for IT help and guidance and
solving “special” hardware problems, to Reinhild Pätzmann for uncomplicated help in
many situations, to Rudolph Böttner for helpful advices, and the whole physics
department for a great cooperation.

I am happy that I had the chance to be part of a cooperation with the Optical
Techniques group at University of Enschede, Twente, in the project of realizing a
photonic wire. Particularly, I am very thankful to Gabriel Sánchez-Mosteiro, Niek van
Hulst and Maria García-Parajó.

I am grateful to Felix Koberling and colleagues at PicoQuant GmbH for the possibility
to use the MicroTime set-up and the opportunity to realize time-resolved
measurements on photonic wires.


For providing custom made analysis software and support in technical problems, I
thank Christian M. Roth and Dirk-Peter Herten.

For proofreading of this manuscript, I am grateful to Sören Doose, Hannes
Neuweiler, Christian Roth, Philip Tinnefeld and Markus Sauer.

Special thanks to Jude Przyborski who was indispensable for revisions of the
manuscript in terms of language questions.

For motivation and support during the last years and always being in reach for my
thoughts, I want to thank Susanne Nessler.

There are many friends who never saw a single molecule, but, they were the perfect
counterpart to scientific work: Sascha Schneider, Stephan Straub, Patrick Knorr,
Judith Pfahler, Jude Przyborski, Martin Wenke, Peter Soba. Thank you all!

My greatest thanks are to my family.


ghkgkyyt ghkgkyyt